# Method for objectively evaluating quality of stereo image

Imported: 13 Feb '17 | Published: 10 Feb '15

USPTO - Utility Patents

## Abstract

A method for objectively evaluating quality of a stereo image is provided. The method obtains a cyclopean image of a stereo image formed in the human visual system by simulating a process that the human visual system deals with the stereo image. The cyclopean image includes three areas: an occlusion area, a binocular fusion area and a binocular suppression area. Representing characteristics of the image according to the singular value of the image has a strong stability. According characteristics of different areas of the human visual system while dealing with the cyclopean image, the distortion degree of the cyclopean image corresponding to the testing stereo image is presented by the singular value distance between cyclopean images respectively corresponding to the testing stereo image and the reference stereo image, in such a manner that an overall visual quality of the testing stereo image is finally evaluated.

## Description

### BACKGROUND OF THE PRESENT INVENTION

1. Field of Invention

The present invention relates to a technique for evaluating stereo image quality, and more particularly to a method for objectively evaluating stereo image quality based on binocular fusion and binocular suppression.

2. Description of Related Arts

With the progressive development of three-dimensional display technologies, stereo image technologies are widely utilized in various fields. However, how to establish an effective method for evaluating stereo image quality which is in accordance with visual perception of human beings has been a hot topic of research in fields of stereo image processing or compression.

From the point of methodology, the stereo image quality evaluation can be classified as two methods, which are respectively a subjective quality-evaluation method and an objective quality-evaluation method. The subjective quality-evaluation method means processing visual perception and quality evaluation on stereo images by a large number of testers. Since the subjective quality-evaluation method is very inconvenient to operate and consumes plenty of time, manpower and physical resources, the ultimate object of stereo image quality assessment is to establish an objective quality-evaluation model which is capable of automatically evaluating and has an evaluation result in accordance with the subjective quality-evaluation result.

The objective quality-evaluation method predicts visual quality of the stereo image to be tested by modeling distortion information, and thus plays an important role in fields of video processing. For example, the objective quality-evaluation method is utilized for monitoring and regulating image quality, optimizing algorithm and parameter settings, evaluating performances of system or algorithm and etc. Currently, multiple methods for objectively evaluating quality of a stereo image are presented by researchers. A first representative method thereof is to evaluate left-view and right-view images of the stereo image respectively by a 2D quality evaluation method directly, and an average value of the results obtained serves as a predictive value for the quality of the stereo image. A second representative method thereof is to separate the stereo-image quality evaluation into two parts: 2D distortion evaluation and stereo distortion evaluation. In the first method, the stereo image is regarded as an assembly of left-view image and right-view image which are independently existed completely without any consideration for the special characteristics of the stereo image. Based on the first method, the second method evaluates the specific quality distortion of the stereo image simply by depth or disparity information thereof. It is obvious that both of the two methods mentioned above are extension or improvement of the 2D-image quality evaluation method. Compared with a 2D image, the stereo image not simply has an extra channel added, and quality evaluation thereof has several difficult problems as following:

• (1) the problem of evaluating the specific distortion type of the stereo image, such as crosstalk, cardboard and keystone;
• (2) the problem of presenting depth perception information of the stereo image and evaluating distortion thereof;
• (3) the problem of interaction between the left-view and the right-view of a stereo image.

Currently, the problems mentioned above have not been solved yet, which leads to inconsistency between the visual quality of the stereo image and evaluating results of the conventional objective quality-evaluation model.

### SUMMARY OF THE PRESENT INVENTION

A technical object of the present invention is to provide a method for objectively evaluating quality of a stereo image based on binocular fusion and binocular suppression, which is capable of objectively reflecting visual quality variation of the stereo image influenced by various methods of image processing and compressing, so as to be consistent with characteristics of human visual system.

In order to solve the technical problems mentioned above, technical solutions utilized by the present invention are as following. A method for objectively evaluating quality of a stereo image comprises following steps of:

{circle around (1)} defining an original stereo image which is utilized for referring in an objective quality evaluation of the stereo image as a reference stereo image, a left-view of the reference stereo image is denoted as Il, a right-view of the reference stereo image is denoted as Ir, a cyclopean image of the reference stereo image formed in a human visual system is denoted as C;

defining a distortion image of the reference stereo image which is used for testing in the objective quality evaluation of the stereo image as a testing stereo image, a left-view of the testing stereo image is denoted as Îl, a right-view of the testing stereo image is denoted as Îr, a cyclopean image of the testing stereo image formed in the human visual system is denoted as Ĉ, wherein sizes of the left-view of the reference stereo image Il, the right-view of the reference stereo image Ir the left-view of the testing stereo image Îr, and the right-view of the testing stereo image Îr are all denoted as W×H, wherein W refers to a width thereof, and H refers to a height thereof;

{circle around (2)} classifying the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three areas, wherein:

a first area thereof is an occlusion area, which is denoted as {circumflex over (R)}occ, information of the occlusion area {circumflex over (R)}occ exists only in the left-view or the right-view of the testing stereo image;

a second area thereof is a binocular suppression area, which is denoted as {circumflex over (R)}bs, information of the binocular suppression area {circumflex over (R)}bs is binocular information having great image content difference or great disparity between the left-view and the right-view of the testing stereo image; and

a third area thereof is a binocular fusion area, which is denoted as {circumflex over (R)}bf, information of the binocular fusion area {circumflex over (R)}bf is binocular information having similar image content and small disparity between the left-view and the right-view of the testing stereo image;

{circle around (3)} dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,

classifying all of the image blocks into three types: an occlusion block, a binocular suppression block and a binocular fusion block,

updating {circumflex over (R)}occ, {circumflex over (R)}bs and {circumflex over (R)}bf according to types of the image blocks, so as to obtain an updated occlusion area {circumflex over (R)}occ′, an updated binocular suppression area {circumflex over (R)}bs′ and an updated binocular fusion area {circumflex over (R)}bf′,

in the cyclopean image C of the reference stereo image formed in the human visual system, obtaining an area Rocc′ corresponding to a position of {circumflex over (R)}occ′, an area Rbs′ corresponding to a position of {circumflex over (R)}bs′, and an area Rbf′ corresponding to a position of {circumflex over (R)}bf′,

processing singular value decomposition respectively on all of the image blocks having a size of k×k in the area {circumflex over (R)}occ′, an area {circumflex over (R)}occ′, corresponding to left-view of it {circumflex over (R)}bs′, an area {circumflex over (R)}bsr′ corresponding to right-view of {circumflex over (R)}bs′, an area {circumflex over (R)}bfl′ corresponding to left-view of {circumflex over (R)}bf′, an area {circumflex over (R)}bfr′ corresponding to right-view of {circumflex over (R)}bf′, the area Rocc′, an area Rbsl′ corresponding to left-view of Rbs′, an area Rbsr′ corresponding to right-view of Rbs′, an area Rbfl′ corresponding to left-view of Rbf′, and an area Rbfr′ corresponding to right-view of Rbf′, in such a manner that singular values respectively corresponding to each image block are obtained, wherein k>1;

{circle around (4)} obtaining an overall distortion degree Qocc of {circumflex over (R)}occ′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}occ′ and Rocc′ and have a size of k×k,

obtaining an overall distortion degree Qbs of {circumflex over (R)}bs′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsl′ and Rbsl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsr′ and Rbsr′ and have a size of k×k, and

obtaining an overall distortion degree Qbf of {circumflex over (R)}bf′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bfl′ and Rbfl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have a size of k×k; and

{circle around (5)} according to the overall distortion degree Qocc of {circumflex over (R)}occ, the overall distortion degree Qbs of {circumflex over (R)}bs′ and the overall distortion degree Qbf of {circumflex over (R)}bf′, calculating an overall distortion degree Q of the testing stereo image by an equation of linear regression, wherein Q=a×Qocc+b×Qbs+c×Qbf, wherein a, b and c are all coefficients and satisfy:

${ a + b + c = 1 0 ≤ a ≤ 1 0 ≤ b ≤ 1 0 ≤ c ≤ 1 ;$

wherein a specific process of the step {circle around (2)} mentioned above comprises:

{circle around (2)}-1 denoting a pixel having a position coordinate of (x,y) in the left-view image Il of the reference stereo image as px,yl, denoting a pixel having a position coordinate of (s,t) in the right-view image Ir of the reference stereo image as ps,tr, denoting a pixel having a position coordinate of (x,y) in the left-view image Îl of the testing stereo image as and denoting a pixel having a position coordinate of (s,t) in the right-view image Îr of the testing stereo image as {circumflex over (p)}s,tr, wherein 1≦x≦W, 1≦y≦H, 1≦s≦W and 1≦t≦H;

{circle around (2)}-2 processing stereo matching on the reference stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the reference stereo image, wherein a specific process thereof comprises:

judging whether px,yl is matched with ps,tr, wherein:

if px,yl is matched with ps,tr, matching relationship of px,yl, and ps,tr is presented as (px,yl, ps,tr), horizontal disparities of px,yl, and ps,tr are both denoted as dx,yh, wherein dx,yh=|s−x|, vertical disparities of px,yl and ps,tr are both denoted as dx,yv, wherein dx,yv=|t−y|,

if no pixel is matched with px,yl in the right-view Ir of the reference stereo image, matching relationship of px,yl is presented as (px,yl,φ), and the horizontal disparity and the vertical disparity of px,yl are both determined to be 255, and

if no pixel is matched with ps,tr in the left-view image Ir of the reference stereo image, matching relationship of px,yl is presented as (px,yl,φ), and the horizontal disparity and the vertical disparity of px,yl are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign;

{circle around (2)}-3 processing stereo matching on the testing stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the testing stereo image, wherein a specific process thereof comprising:

judging whether {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr wherein:

if pxyl is matched with {circumflex over (p)}s,tr matching relationship of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr is presented as ({circumflex over (p)}x,yl, {circumflex over (p)}s,tr), the horizontal disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yh, wherein a {circumflex over (d)}x,yh=|s−x| the vertical disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yv, wherein {circumflex over (d)}x,yv=|t−y|,

if no pixel is matched with {circumflex over (p)}x,yl in the right-view Îr of the testing stereo image, matching relationship of {circumflex over (p)}x,yl is presented as ({circumflex over (p)}x,yl,φ), and the horizontal disparity and the vertical disparity of {circumflex over (p)}x,yl are both determined to be 255, and

if no pixel is matched with {circumflex over (p)}x,yl in the left-view Îr of the testing stereo image, matching relationship of {circumflex over (p)}s,tr is presented as (φ,{circumflex over (p)}s,tr), and the horizontal disparity and the vertical disparity of {circumflex over (p)}s,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign;

{circle around (2)}-4 dividing the cyclopean image of the testing stereo image formed in the human visual system Ĉinto an occlusion area {circumflex over (R)}occ and a match area {circumflex over (R)}match, wherein the occlusion area {circumflex over (R)}occ comprises a left-view occlusion area {circumflex over (R)}occl and a right-view occlusion area {circumflex over (R)}occr, wherein {circumflex over (R)}occ={circumflex over (R)}occl∪{circumflex over (R)}occr,

wherein the left-view occlusion area {circumflex over (R)}occl is a set of pixels having a matching relationship of ({circumflex over (p)}x,yl,φ) in the left-view Îl of the testing stereo image, which are presented as {circumflex over (R)}occl={{circumflex over (p)}x,yl|({circumflex over (p)}x,yl,φ){circumflex over (p)}x,ylεÎl},

the right-view occlusion area {circumflex over (R)}occr is a set of pixels having a matching relationship of (φ, {circumflex over (p)}s,tr) in the right-view Îr of the testing stereo image, which is presented as {circumflex over (R)}occr={{circumflex over (p)}s,tr|φ,{circumflex over (p)}s,tr){circumflex over (p)}s,trεÎr},

the match area {circumflex over (R)}match is a set of matched pixels of the left-view Îl of the testing stereo image and the right-view Îr of the testing stereo image, which has a matching relationship of ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr), wherein “∪” is an operational sign of a union, “” is an operational sign of “and”; and

{circle around (2)}-5 further dividing the match area {circumflex over (R)}match into a binocular suppression area {circumflex over (R)}bs and a binocular fusion area {circumflex over (R)}bf, wherein:

the binocular suppression area {circumflex over (R)}bs is an integrated area formed by processing binocular suppression on the left-view corresponding area {circumflex over (R)}bsl of {circumflex over (R)}bs and the right-view corresponding area {circumflex over (R)}bsr of {circumflex over (R)}bs, and

the binocular fusion area {circumflex over (R)}bf is an integrated area formed by processing binocular fusion on the left-view corresponding area {circumflex over (R)}bfl of {circumflex over (R)}bf and the right-view corresponding view {circumflex over (R)}bf of {circumflex over (R)}bf, wherein:
{circumflex over (R)}bsl={{circumflex over (p)}x,yl|{circumflex over (p)}x,yl|εÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},
{circumflex over (R)}bsr={{circumflex over (p)}x,yr|{circumflex over (p)}x,yl|εÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},
{circumflex over (R)}bfr={{circumflex over (p)}s,tr|{circumflex over (p)}x,yl|εÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}, and
{circumflex over (R)}bfr={{circumflex over (p)}s,tr|{circumflex over (p)}x,yl|εÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|};

wherein specific process of the step {circle around (3)} mentioned above comprises:

{circle around (3)}-1 dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,

classifying all of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three types: an occlusion block, a binocular suppression block and a binocular fusion block, wherein specific process of the classifying comprises:

denoting any one of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as {circumflex over (B)}, and

judging whether there is a pixel belonging to {circumflex over (R)}occ in the {circumflex over (B)}, wherein if yes, mark {circumflex over (B)} as a occlusion block; if no, further judge whether there is a pixel belonging to {circumflex over (R)}bs in {circumflex over (B)}, if yes, mark {circumflex over (B)} as a binocular suppression block, if no mark {circumflex over (B)} as a binocular fusion block, wherein k>1;

{circle around (3)}-2 defining an area constituted by all of the occlusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated occlusion area, which is denoted as {circumflex over (R)}occ′,

defining an area constituted by all of the binocular suppression blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular suppression area, which is denoted as {circumflex over (R)}bs′, and

defining an area constituted by all of the binocular fusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular fusion area, which is denoted as {circumflex over (R)}bf′,

wherein the left-view corresponding area of the updated occlusion area {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occl, the right-view corresponding area of {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occr′, the left-view corresponding area of the updated binocular suppression area {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the right-view corresponding area of {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsr′, the left-view corresponding area of the updated binocular fusion area {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfl′, and the right-view corresponding area of {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfr′;

{circle around (3)}-3 denoting the area corresponding to the position of {circumflex over (R)}occ′ in the cyclopean image C of the reference stereo image formed in the human visual system as {circumflex over (R)}occ′, wherein Rocc′=Roccl′∪Roccr′, wherein “∪” is an operational sign for union in a set,

wherein Roccl′ is the area corresponding to the position of {circumflex over (R)}occl′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}occl′},

wherein Roccr′ is the area corresponding to the position of {circumflex over (R)}occr in the cyclopean image C of the reference stereo image formed in the human visual system, Roccr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}occr′};

{circle around (3)}-4 denoting the area corresponding to the position of {circumflex over (R)}bsl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rb,sl′, wherein Rb,sl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bsl′}, and

denoting the area corresponding to the position of {circumflex over (R)}b,sl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsr′, wherein Rbsr′={ps,tr|ps,trεIrps,trε{circumflex over (R)}bsr′};

{circle around (3)}-5 denoting the area corresponding to the position of {circumflex over (R)}bfl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfl′, wherein Rbfl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bfl′}, and

denoting the area corresponding to the position of {circumflex over (R)}bfr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfr′, wherein Rbfr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bfr′}; and

{circle around (3)}-6 processing the singular value decomposition respectively on all of the image blocks having the size of k×k in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′, in such a manner that the singular values respectively corresponding to all image blocks having the size of k×k and in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ are obtained,

wherein in the step {circle around (3)}-1, k=4;

wherein specific process of the step {circle around (4)} mentioned above comprises:

{circle around (4)}-1 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}occ′ and Rocc′ and have the size of k×k, wherein an ith occlusion block in {circumflex over (R)}occ′ is denoted as {circumflex over (B)}i, in Rocc′ an ith occlusion block in a corresponding position of {circumflex over (B)}i is denoted as Bi, a singular value distance between {circumflex over (B)}i and Bi is calculated and is denoted as Docc(i),

$D occ ⁡ ( i ) = ∑ j = 1 k ⁢ ⁢ ( s i , j - s ^ i , j ) 2 ,$

calculating the singular value distances corresponding to all of the occlusion blocks in {circumflex over (R)}occ′ as the overall distortion degree of {circumflex over (R)}occ′, which is denoted as Qocc, wherein

$Q occ = 1 N occ ⁢ ∑ i = 1 N occ ⁢ ⁢  D occ ⁡ ( i ) - D occ ⁡ ( m )  ,$

wherein 1≦i≦Nocc, Nocc refers to a number of all of the occlusion blocks in {circumflex over (R)}occ′, and Nocc is also a number of all of the occlusion blocks in Rocc′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si,j refers to a j th singular value of Bi, ŝi,j refers to the jth singular value of {circumflex over (B)}i, “∥” is an absolute value sign, and Docc(m) refers to a median of Docc(1), Docc(2), . . . , Docc(Nocc);

{circle around (4)}-2 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsl′ and Rbsl′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsl′ is denoted as {circumflex over (B)}i′l, in Rbsl′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′l is denoted as Bi′l, a singular value distance between {circumflex over (B)}i′l and Bi′l is calculated and is denoted as Dbsl(i′),

$D bs l ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ⁢ ( s i ′ , j l - s ^ i ′ , j l ) 2 ,$

calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}b,sr′ and Rbsr′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsr′ is denoted as Bi′r, in Rbsr′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′r is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r and Bi′r is calculated and is denoted as Dbsr(i′),

$D bs r ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ⁢ ( s i ′ , j r - s ^ i ′ , j r ) 2 ,$

calculating the singular value distances of all of the binocular suppression blocks in {circumflex over (R)}bs′, wherein an i′th binocular suppression block in {circumflex over (R)}bs′ is denoted as {circumflex over (B)}i′, and singular value distance thereof is calculated and denoted as Dbs(i′), Dbs(i′)=min {Dbsl(i′), Dbsr(i′)}, and

calculating the singular value distances corresponding to all of the binocular suppression blocks in {circumflex over (R)}bs′ as the overall distortion degree of {circumflex over (R)}bs′, which is denoted as Qbs, wherein

$Q bs = 1 N bs ⁢ ∑ i ′ = 1 N bs ⁢ ⁢  D bs ⁡ ( i ′ ) - D bs ⁡ ( m )  ,$
wherein 1≦i′≦Nbs, Nbs refers to a number of all of the binocular suppression blocks in {circumflex over (R)}bsl′ or {circumflex over (R)}bsr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si′,jl refers to a jth singular value of Bi′l, ŝi′,jl refers to the jth singular value of {circumflex over (B)}i′l, si′,jr refers to a j th singular value of Bi′r, ŝi′,jr refers to a jth singular value of {circumflex over (B)}i′r, min ( ) is a sign for a minimum value, “∥” is an absolute value sign, and Dbs(m) refers to a median of Dbs(1), Dbs(2), . . . , Dbs(Nbs); and

{circle around (4)}-3 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfl′ and Rbfl′ and have the size of k×k, wherein an i′th binocular fusion block in {circumflex over (R)}bfl′ is denoted as {circumflex over (B)}i′l, in Rbfl′ an i′th binocular fusion block in a corresponding position of {circumflex over (B)}i″l is denoted as Bi″l, a singular value distance between {circumflex over (B)}i″l and Bi″l is calculated and is denoted as Dbfl(i″)

$D bf l ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ⁢ ( s i ″ , j l - s ^ i ″ , j l ) 2 ,$

calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfr′ is denoted as {circumflex over (B)}i′r, in Rbfr′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″r is denoted as Bi″r, a singular value distance between {circumflex over (B)}i′r, and Bi″r is calculated and is denoted as Dbfr(i″),

$D bf r ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ⁢ ( s i ″ , j r - s ^ i ″ , j r ) 2 ,$

calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfl′ for serving as an overall distortion degree of {circumflex over (R)}bfl′, which is denoted as Qbfl,

$Q bf l = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢ ⁢  D bf l ⁡ ( i ″ ) - D bf l ⁡ ( m )  ,$

calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfr′ for serving as an overall distortion degree of {circumflex over (R)}bfr′, which is denoted as Qbfr,

$Q bf r = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢ ⁢  D bf r ⁡ ( i ″ ) - D bf r ⁡ ( m )  ,$
and

calculating an overall distortion degree of {circumflex over (R)}bf′ according to Qbfl and Qbfr, which is denoted as Qbf, Qbf=0.7×(Qbfl+Qbfr),

wherein 1≦i″≦Nbf, Nbf refers to a number of all of the binocular fusion blocks in {circumflex over (R)}bfl′ or {circumflex over (R)}bfr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si″,jl refers to a j th singular value of Bil, ŝi″,jl refers to the jth singular value of {circumflex over (B)}i″l, si″,jr refers to a j th singular value of Bi″r, si″,jr refers to a j th singular value of {circumflex over (B)}i″r, “∥” is an absolute value sign, and Dbfl(m) refers to a median of Dbfl(1), Dbfl(2), . . . , Dbfl(Nbf), Dbfr(m) refers to a median of Dbfr(1), Dbfr(2), . . . , Dbfr(Nbf);

wherein in the step {circle around (5)}, a=0, b=0.5, c=0.5.

Compared with the prior art, advantage of the present invention lies in obtaining the cyclopean image of the stereo image formed in the human visual system by simulating the process that the human visual system deals with the stereo image, wherein the cyclopean image comprises three areas: the occlusion area, the binocular fusion area and the binocular suppression area. Representing characteristics of the image according to the singular value of the image has a strong stability. According characteristics of different areas of the human visual system while dealing with the cyclopean image, the distortion degree of the cyclopean image corresponding to the testing stereo image is presented by the singular value distance between cyclopean images respectively corresponding to the testing stereo image and the reference stereo image, in such a manner that an overall visual quality of the testing stereo image is finally evaluated. The method for objectively evaluating quality of the stereo image of the present invention is capable of objectively reflecting visual quality variations of the stereo image under influence of various image processing or suppressing methods. Furthermore, evaluation performance of the method of the present invention is not influenced by contents and distortion types of the stereo image, and is in accordance with subjective visual perception of human eyes.

These and other objectives, features, and advantages of the present invention will become apparent from the following detailed description, the accompanying drawings, and the appended claims.

### DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT

Combined with the accompanying drawings and the preferred embodiment, further and detailed description of the present invention is illustrated as following.

Studies of human visual physiology show that an ocular distance of human beings is about 60˜65 mm. The distance causes that two eyes are in a slightly different location and angle while watching a certain scene, which further causes that the scene has slightly different images formed between retinal of the two eyes. However, in practice, people having normal vision do not feel that the two images formed in the two eyes are overlapped while watching the scene. Image formed in human brain is a stereo cyclopean image, i.e., single-eyed view image. This phenomenon is a result of binocular fusion and binocular suppression processed by human visual system. In the human visual system, due to existing of a Panum area, the binocular fusion is occurred, areas respectively corresponding to retinal of left eye and right eye are not necessary to accurately locate at a same position, and a slight disparity is allowed thereof. However, if great content difference or great disparity is existed between images respectively corresponding to retinal of the left eye and the right eye, the human visual system is not capable of processing the binocular fusion on conflict information of two eyes. Then phenomenon of diplopia or binocular visual confusion appears, and a binocular rivalry status starts. However, the normal human visual system can not endure the binocular rivalry status for a long time, and the binocular rivalry status is finally transferred to a status of binocular suppression processing. Two situations may occur in the binocular suppression processing:

if contents of the images in the areas respectively corresponding to the retinal of the left eye and the right eye have great differences, information of the retinal of the left eye and the right eye will intermittently suppress an opposite side in turn, i.e., in the areas respectively corresponding to the retinal of the left eye and the right eye, image information of the retinal of the left eye and the right eye is intermittently shown in turn; and

if contents of the images in the areas respectively corresponding to the retinal of the left eye and the right eye are relatively similar but have great disparity therebetween, in the area, information of one eye on the retinal of the left eye or the right eye will enduringly suppress thereof the other eye. Usually, the image having a more clear contour in one eye suppresses thereof the other eye.

A process of the binocular fusion and the binocular suppression in the human visual system is shown as in FIG. 1 of the drawings.

According to characteristic of binocular stereo vision in human visual system, the present invention provides a method for objectively evaluating quality of a stereo image based on the binocular fusion and the binocular suppression. The method for objectively evaluating quality of the stereo image of the present invention firstly simulates a process that the human visual system treats the stereo image, so as to obtain the cyclopean image formed in human brain by the stereo image, then processes mathematical modeling on the cyclopean image, evaluates distortion degree of the cyclopean image by a full reference quality evaluation method, in such a manner that overall visual quality of the stereo image is finally obtained.

As shown in FIG. 2 which is an overall flow block diagram, the present invention provides a method for objectively evaluating quality of a stereo image based on binocular fusion and binocular suppression, comprising steps as following.

{circle around (1)} Defining an original stereo image which is utilized for referring in an objective quality evaluation of the stereo image as a reference stereo image, a left-view of the reference stereo image is denoted as Il, a right-view of the reference stereo image is denoted as Ir, a cyclopean image of the reference stereo image formed in a human visual system is denoted as C.

Defining a distortion image of the reference stereo image which is used for testing in the objective quality evaluation of the stereo image as a testing stereo image, a left-view of the testing stereo image is denoted as Îl, a right-view of the testing stereo image is denoted as Îr, a cyclopean image of the testing stereo image formed in the human visual system is denoted as Ĉ, wherein sizes of the left-view of the reference stereo image Il, the right-view of the reference stereo image Ir the left-view of the testing stereo image Îl, and the right-view of the testing stereo image Îr are all denoted as W×H, wherein W refers to a width thereof, and H refers to a height thereof.

{circle around (2)} According to characteristics of the human visual system processing stereo image, classifying the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three areas.

A first area thereof is an occlusion area, which is denoted as {circumflex over (R)}occ, information of the occlusion area {circumflex over (R)}occ exists only in the left-view or the right-view of the testing stereo image.

A second area thereof is a binocular suppression area, which is denoted as {circumflex over (R)}bs information of the binocular suppression area {circumflex over (R)}bs is binocular information having great image content difference or great disparity between the left-view and the right-view of the testing stereo image, i.e., the binocular suppression area is an area produced by binocular suppression processing of the human visual system, in which the corresponding image contents in the left-view and right-view of the stereo image have great difference or great disparities.

A third area thereof is a binocular fusion area, which is denoted as {circumflex over (R)}bf, information of the binocular fusion area {circumflex over (R)}bf is binocular information having similar image content and small disparity between the left-view and the right-view of the testing stereo image, i.e. the binocular fusion area is an area produced by binocular fusion processing of the human visual system, in which the corresponding image contents in the left-view and right-view of the stereo image are similar with each other and the corresponding disparities thereof are small.

In this preferred embodiment, a specific process of the step {circle around (2)} mentioned above comprises:

{circle around (2)}-1 Denoting a pixel having a position coordinate of (x,y) in the left-view image Il of the reference stereo image as px,yl, denoting a pixel having a position coordinate of (s,t) in the right-view image Ir of the reference stereo image as ps,tr, denoting a pixel having a position coordinate of (x,y) in the left-view image Îl of the testing stereo image as {circumflex over (p)}x,yl, and denoting a pixel having a position coordinate of (s,t) in the right-view image Îr of the testing stereo image as {circumflex over (p)}s,tr, wherein 1≦x≦W, 1≦y≦H, 1≦s≦W and 1≦t≦H.

{circle around (2)}-2 Processing stereo matching on the reference stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the reference stereo image, wherein a specific process thereof comprises:

judging whether px,yl is matched with ps,tr, wherein:

if px,yl is matched with ps,tr, matching relationship of px,yl and ps,tr is presented as (px,yl,ps,tr), the horizontal disparities of px,yl and ps,tr are both denoted as dx,yh, wherein dx,yh=|s−x|, the vertical disparities of px,yl and ps,tr are both denoted as dx,yv, wherein dx,yv=|t−y|,

if no pixel is matched with px,yl in the right-view Ir of the reference stereo image, matching relationship of px,yl is presented as (px,yl,φ), and the horizontal disparity and the vertical disparity of px,yl are both determined to be 255, and

if no pixel is matched with ps,tr in the left-view image Il of the reference stereo image, matching relationship of ps,tr is presented as (φ, ps,tr), and the horizontal disparity and the vertical disparity of ps,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign.

{circle around (2)}-3 Processing stereo matching on the testing stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the testing stereo image, wherein a specific process thereof comprising:

judging whether {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr, wherein:

if {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr matching relationship of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr is presented as ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr) the horizontal disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as dx,yh, wherein a {circumflex over (d)}x,yh=|s−x|, the vertical disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yv, wherein {circumflex over (d)}x,yv=|t−y|,

if no pixel is matched with {circumflex over (p)}x,yl in the right-view Îr of the testing stereo image, matching relationship of {circumflex over (p)}x,yl is presented as ({circumflex over (p)}x,yl,φ), and the horizontal disparity and the vertical disparity of {circumflex over (p)}x,yl are both determined to be 255, and

if no pixel is matched with {circumflex over (p)}s,tr, in the left-view Îl of the testing stereo image, matching relationship of {circumflex over (p)}s,tr is presented as (φ, {circumflex over (p)}s,tr), and the horizontal disparity and the vertical disparity of {circumflex over (p)}s,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign.

{circle around (2)}-4 After stereo matching, according to characteristics of the human visual system processing stereo image, dividing the cyclopean image of the testing stereo image formed in the human visual system Ĉ into an occlusion area {circumflex over (R)}occ and a match area {circumflex over (R)}match, the occlusion area {circumflex over (R)}occ is a set of pixels of the left-view and right-view of the testing stereo image which fail to match themselves in another view of the testing stereo image, thus the occlusion area {circumflex over (R)}occ comprises a left-view occlusion area {circumflex over (R)}occl and a right-view occlusion area {circumflex over (R)}occr, wherein {circumflex over (R)}occ={circumflex over (R)}occl∪{circumflex over (R)}occr,

wherein the left-view occlusion area {circumflex over (R)}occl is a set of pixels having a matching relationship of ({circumflex over (p)}x,yl,φ) in the left-view Îl of the testing stereo image, which are presented as {circumflex over (R)}occl={{circumflex over (p)}x,yl|({circumflex over (p)}x,yl,φ){circumflex over (p)}x,ylεÎl},

the right-view occlusion area {circumflex over (R)}occr is a set of pixels having a matching relationship of (φ, {circumflex over (p)}s,tr) in the right-view Îr of the testing stereo image, which is presented as {circumflex over (R)}occr={{circumflex over (p)}s,tr|(φ, {circumflex over (p)}s,tr){circumflex over (p)}s,trεÎr},

the match area {circumflex over (R)}match is a set of matched pixels of the left-view Îl of the testing stereo image and the right-view Îr of the testing stereo image, which having a matching relationship of ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr), wherein “∪” is an operational sign of a union, “” is an operational sign of “and”.

{circle around (2)}-5 Further dividing the match area {circumflex over (R)}match into a binocular suppression area {circumflex over (R)}bs and a binocular fusion area {circumflex over (R)}bf.

Binocular suppression area {circumflex over (R)}bs is an area of which the disparity between the left-view and right-view of the original stereo image becomes larger after distortion process. The binocular suppression area {circumflex over (R)}bs is an integrated area formed by processing binocular suppression on the left-view corresponding area {circumflex over (R)}bsl of {circumflex over (R)}bs and the right-view corresponding area {circumflex over (R)}bsr of {circumflex over (R)}bs, that is, {circumflex over (R)}bs=Sup({circumflex over (R)}bsl,{circumflex over (R)}bsr), Sup( ) denotes binocular suppression processing of the human visual system, which shows that the information of a view with relatively good quality suppresses the information of another view with relatively poor quality, and
{circumflex over (R)}bsl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎ|({circumflex over (p)}x,yl,{circumflex over (p)}s,tr){circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},
{circumflex over (R)}bsr={{circumflex over (p)}s,tlεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|}.

Binocular fusion area {circumflex over (R)}bf is an area of which the disparity between the left-view and right-view of the original stereo image is not changed or becomes smaller after distortion process. The binocular fusion area {circumflex over (R)}bf is an integrated area formed by processing binocular fusion on the left-view corresponding area {circumflex over (R)}bf of {circumflex over (R)}bf and the right-view corresponding view {circumflex over (R)}bfr of {circumflex over (R)}bf, that is, {circumflex over (R)}bf=Fus{{circumflex over (R)}bfl,{circumflex over (R)}bfr}, Fus( ) denotes binocular fusion processing and binocular superposition processing of human visual system, and
{circumflex over (R)}bfl={{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|},
{circumflex over (R)}bfr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|}.

FIG. 3 shows a sketch view of an area classifying result of Akko & Kayo. The original stereo image Akko & Kayo is used as the reference stereo image in this embodiment, and the corresponding testing stereo image is distorted by Gaussian blurring. PSNR of the left-view of the testing stereo image is 24.450 dB, and PSNR of the right-view of the testing stereo image is 23.862 dB.

In this embodiment, the stereo matching of a reference stereo image or a testing stereo image is implemented with a pixel-based software of MATCH v3.3 provided by Cornell University.

{circle around (3)} Dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,

classifying all of the image blocks into three types: an occlusion block, a binocular suppression block and a binocular fusion block,

updating {circumflex over (R)}occ, {circumflex over (R)}bs and {circumflex over (R)}bf according to types of the image blocks, so as to obtain an updated occlusion area {circumflex over (R)}occ′, an updated binocular suppression area {circumflex over (R)}bs′ and an updated binocular fusion area {circumflex over (R)}bf′,

in the cyclopean image C of the reference stereo image formed in the human visual system, obtaining an area Rocc′ corresponding to a position of {circumflex over (R)}occ′, an area Rbs′ corresponding to a position of {circumflex over (R)}bs′, and an area Rbf′ corresponding to a position of {circumflex over (R)}bf′,

processing singular value decomposition respectively on all of the image blocks having a size of k×k in the area {circumflex over (R)}occ′, an area {circumflex over (R)}bsl′ corresponding to left-view of {circumflex over (R)}bsl′, an area {circumflex over (R)}bsr′ corresponding to right-view of {circumflex over (R)}bs′, an area {circumflex over (R)}bfl′ corresponding to left-view of {circumflex over (R)}bf′, an area {circumflex over (R)}bfr′ corresponding to right-view of {circumflex over (R)}bf′, the area Rocc′, an area Rbsl′ corresponding to left-view of Rbs′, an area Rbsr′ corresponding to right-view of Rbs′, an area Rbfl′ corresponding to left-view of Rbf′, and an area Rbfr′ corresponding to right-view of Rbf′, in such a manner that singular values respectively corresponding to each image block are obtained, wherein k>1.

A specific process of the step {circle around (3)} mentioned above comprises:

{circle around (3)}-1 dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,

classifying all of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three types: an occlusion block, a binocular suppression block and a binocular fusion block, wherein specific process of the classifying comprises:

denoting any one of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as {circumflex over (B)}, and

judging whether there is a pixel belonging to {circumflex over (R)}occ in the {circumflex over (B)}, wherein if yes, mark {circumflex over (B)} as a occlusion block; if no, further judge whether there is a pixel belonging to {circumflex over (R)}bs in {circumflex over (B)}, if yes, mark {circumflex over (B)} as a binocular suppression block, if no mark {circumflex over (B)} as a binocular fusion block, wherein k>1, and in this embodiment k=4;

{circle around (3)}-2 defining an area constituted by all of the occlusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated occlusion area, which is denoted as {circumflex over (R)}occ′,

defining an area constituted by all of the binocular suppression blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular suppression area, which is denoted as {circumflex over (R)}bs′, and

defining an area constituted by all of the binocular fusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular fusion area, which is denoted as {circumflex over (R)}bf′,

wherein the left-view corresponding area of the updated occlusion area {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occl′, the right-view corresponding area of {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occr′, the left-view corresponding area of the updated binocular suppression area {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the right-view corresponding area of {circumflex over (R)}bs′ is denoted as {circumflex over (R)}b,sr′, the left-view corresponding area of the updated binocular fusion area {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfl′, and the right-view corresponding area of {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfr′;

{circle around (3)}-3 denoting the area corresponding to the position of {circumflex over (R)}occ in the cyclopean image C of the reference stereo image formed in the human visual system as Rocc′, wherein Rocc′=Roccl′∪Roccr′, wherein “∪” is an operational sign for union in a set,

wherein Roccl′ is the area corresponding to the position of {circumflex over (R)}occl′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}occl′},

wherein Roccr′ is the area corresponding to the position of {circumflex over (R)}occr′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}occr′},

{circle around (3)}-4 denoting the area corresponding to the position of {circumflex over (R)}bsl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsl, wherein Rbsl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bsl′}, an

denoting the area corresponding to the position of {circumflex over (R)}b,sr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsr′, wherein Rbsr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bsr′};

{circle around (3)}-5 denoting the area corresponding to the position of {circumflex over (R)}bfl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfl′, wherein Rbfl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bfl′}, and

denoting the area corresponding to the position of {circumflex over (R)}bfr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfr′, wherein Rbfr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bfr′}; and

{circle around (3)}-6 processing the singular value decomposition respectively on all of the image blocks having the size of k×k in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′ and {circumflex over (R)}bfr′, in such a manner that the singular values respectively corresponding to all image blocks having the size of k×k and in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, {circumflex over (R)}occ′, {circumflex over (R)}bsl′, Rbsr′, Rbfl′ and Rbfr′ are obtained. For an image block having a size of k×k, the singular value decomposition can be represented as matrix operation. Let G denote the image block, then the singular value decomposition can be represented as G=USVT, wherein U and V are orthogonal matrices, S is a diagonal matrix, S=diag(s1, s2, . . . , sk), and all of diagonal elements of S are called as singular value of G.

{circle around (4)} Obtaining an overall distortion degree Qocc of {circumflex over (R)}occ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}occ′ and Rocc′ and have a size of k×k,

obtaining an overall distortion degree Qbs of {circumflex over (R)}bs′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsl′ and Rbsl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsr′ and Rbsr′ and have a size of k×k, and

obtaining an overall distortion degree Qbf of {circumflex over (R)}bf′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bfl′ and Rbfl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have a size of k×k.

A specific process of the step {circle around (4)} is as following.

{circle around (4)}-1 Calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}occ′ and Rocc′ and have the size of k×k, wherein an ith occlusion block in {circumflex over (R)}occ′ is denoted as {circumflex over (B)}i, in Rocc′ an ith occlusion block in a corresponding position of {circumflex over (B)}i is denoted as Bi, a singular value distance between {circumflex over (B)}i and Bi is calculated and is denoted as Docc(i),

$D occ ⁡ ( i ) = ∑ j = 1 k ⁢ ⁢ ( s i , j - s ^ i , j ) 2 .$

Calculating the singular value distances corresponding to all of the occlusion blocks in {circumflex over (R)}occ′ as the overall distortion degree of {circumflex over (R)}occ′, which is denoted as Qocc, wherein

$Q occ = 1 N occ ⁢ ∑ i = 1 N occ ⁢ ⁢  D occ ⁡ ( i ) - D occ ⁡ ( m )  ,$
wherein 1≦i≦Nocc, Nocc refers to a number of all of the occlusion blocks in {circumflex over (R)}occ′, and Nocc is also a number of all of the occlusion blocks in Rocc′, 1≦j≦k refers to a number of singular values in one of the image blocks, si,j refers to a j th singular value of Bi, Ŝi,j refers to the jth singular value of {circumflex over (B)}i, “∥” is an absolute value sign, and Docc(m) refers to a median of Docc(1), Docc(2), . . . , Docc(Nocc).

{circle around (4)}-2 Calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsl′ and Rb,sl′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsl′ is denoted as {circumflex over (B)}i′l, in Rb,sl′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′l is denoted as Bi′l, a singular value distance between {circumflex over (B)}i′l and Bi′l is calculated and is denoted as Dbsl(i′),

$D bs l ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ⁢ ( s i ′ , j l - s ^ i ′ , j l ) 2 .$

Calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsr′ and Rbsr′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsr′ is denoted as {circumflex over (B)}i′r, in Rbsr′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′r, is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r, and Bi′r is calculated and is denoted as Dbsr(i′),

$D bs r ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ⁢ ( s i ′ , j r - s ^ i ′ , j r ) 2 .$

Calculating the singular value distances of all of the binocular suppression blocks in {circumflex over (R)}bs′, wherein an i′th binocular suppression block in {circumflex over (R)}bs′ is denoted as {circumflex over (B)}i′, and its singular value distance is calculated and denoted as Dbs(i′). Since usually the image having a more clear contour in one eye suppresses thereof the other eye in binocular suppression process of human visual system, and the singular value distance between a distorted stereo image and its original stereo image reflects the distortion degree of the distorted stereo image, which means that the larger the singular value distance is, the greater the distortion is, thus, the singular value distance Dbs(i′) of {circumflex over (B)}i′ is determined by a smaller one of Dbsl(i′) and Dbsr(i′), that is, Dbs(i′)=min{Dbsl(i′), Dbsr(i′)}.

Calculating the singular value distances corresponding to all of the binocular suppression blocks in {circumflex over (R)}bs′ as the overall distortion degree of {circumflex over (R)}bs′, which is denoted as Qbs, wherein

$Q bs = 1 N bs ⁢ ∑ i ′ = 1 N bs ⁢ ⁢  D bs ⁡ ( i ′ ) - D bs ⁡ ( m )  ,$
wherein 1≦i′≦Nbs, Nbs refers to a number of all of the binocular suppression blocks in {circumflex over (R)}bsl′ or {circumflex over (R)}b,sl′, 1≦j≦k refers to a number of singular values in one of the image blocks, si′,jl refers to a jth singular value of Bi′l, ŝi′,jl refers to the jth singular value of {circumflex over (B)}i′l, si′,jr refers to a j th singular value of Bi′r, ŝi′,jr refers to a jth singular value of {circumflex over (B)}i′r, min ( ) is a sign for a minimum value, “∥” is an absolute value sign, and Dbs(m) refers to a median of Dbs(1), Dbs(2), . . . , Dbs(Nbs).

{circle around (4)}-3 Calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfl′ and {circumflex over (R)}bfl′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfl′ is denoted as {circumflex over (B)}i″l in Rbfl′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″l, is denoted as Bi″l, a singular value distance between {circumflex over (B)}i″l and Bi″l is calculated and is denoted as Dbfl(i″),

$D bf l ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ⁢ ( s i ″ , j l - s ^ i ″ , j l ) 2 .$

• Calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfr′ and {circumflex over (R)}bfr′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfr′ is denoted as {circumflex over (B)}i′r, in Rbfr′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″r is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r, and Bi″r is calculated and is denoted as Dbfr(i″),

$D bf r ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ⁢ ( s i ″ , j r - s ^ i ″ , j r ) 2 .$

Calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfl′ for serving as an overall distortion degree of {circumflex over (R)}bfl′, which is denoted as Qbfl,

$Q bf l = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢ ⁢  D bf l ⁡ ( i ″ ) - D bf l ⁡ ( m )  .$

Calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfr′ for serving as an overall distortion degree of {circumflex over (R)}bfr′ which is denoted as Qbfr,

$Q bf r = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢ ⁢  D bf r ⁡ ( i ″ ) - D bf r ⁡ ( m )  .$

According to characteristic of binocular superposition processing of human visual system, the binocular sensitiveness is 1.4 times thereof the monocular. Therefore, an overall distortion degree of {circumflex over (R)}bf′ is calculated and denoted as Qbf according to Qbfl and Qbfr, Qbf=0.7×(Qbfl+Qbfr).

Here, 1≦i″≦Nbf, Nbf refers to a number of all of the binocular fusion blocks in {circumflex over (R)}bfl′ or {circumflex over (R)}bfr′, 1≦j≦l, k refers to a number of singular values in one of the image blocks, si″,jl refers to a j th singular value of Bi′l, si″,jl refers to the jth singular value of {circumflex over (B)}i″l, si″,jr refers to a j th singular value of Bi″r, ŝi″,jr refers to a j th singular value of {circumflex over (B)}i″r, “∥” is an absolute value sign, and Dbfl(m) refers to a median of Dbfl(1), Dbfl(2), . . . , Dbfl(Nbf), Dbfr(m) refers to a median of Dbfr(1), Dbfr(2), . . . , Dbfr(Nbf).

{circle around (5)} Quality of the testing stereo image are influenced by distortions in the three areas including {circumflex over (R)}occ′, {circumflex over (R)}b′ and {circumflex over (R)}bf′, and the distortions in the three areas are independent with each other. The present invention integrates the overall distortion degrees Qocc, Qbs and Qbf with respect to the three areas according to their contributions to the overall visual quality of the testing stereo image, so as to obtain the overall distortion degree Q of the testing stereo image. According to the overall distortion degree Qocc of {circumflex over (R)}occ′, the overall distortion degree Qbs of {circumflex over (R)}bs′ and the overall distortion degree Qbf of {circumflex over (R)}bf′, calculating an overall distortion degree Q of the testing stereo image by an equation of linear regression, wherein Q=a×Qocc+b×Qbs+c×Qbf, wherein a, b and c are all coefficients and satisfy:

${ a + b + c = 1 0 ≤ a ≤ 1 0 ≤ b ≤ 1 0 ≤ c ≤ 1.$
In this embodiment, a=0, b=0.5, c=0.5.

9 high-resolution color stereo images and corresponding distortion stereo images thereof are selected from testing sequences provided by Mobile3DTV for serving as testing database of the stereo image, so as to measure consistency between the method of the present invention and visual subjective perception of human beings. The testing database of the stereo image comprises the 9 high-resolution color stereo images which is shown in FIG. 4 of the drawings, and 234 distortion stereo images in total which comprise distortion stereo images of the 9 high-resolution color stereo images at 5 distortion levels by gaussian blur, distortion stereo images of the 9 high-definition color stereo images at 5 distortion levels by white Gaussian noise, distortion stereo images of the 9 high-definition color stereo images at 5 distortion levels by JPEG compression, distortion stereo images of the 9 high-definition color stereo images at 5 distortion levels by JPEG-2000 compression, and distortion stereo images of the 9 high-definition color stereo images at 6 distortion levels by H.264 compression.

FIG. 5 is a scatter diagram showing a subjective evaluation result DMOS of the 234 testing stereo images and a predicted value DMOSp thereof obtained by nonlinear processing on a result according to the method of the present invention. In the FIG. 5, the higher an intensity of data points are around a line y=x, the better the consistency is between the result according to the method of the present invention and the subjective evaluation result. In order to illustrate evaluation performance of the method of the present invention better, four performance indexes are utilized for evaluation according to testing standard of video quality evaluation group:

• (1) Pearson linear correlation coefficient (CC), which is utilized for reflecting predication accuracy of an objective evaluation model;
• (2) root-mean-square error (RMSE), which is often utilized for measuring accuracy of the objective evaluation model besides the Pearson linear correlation coefficient;
• (3) Spearman rank correlation coefficient (SROCC), which is utilized for measuring monotonicity of the objective evaluation model; and
• (4) obnormal value ratio (OR), which is utilized for reflecting dispersion degree of the objective evaluation model.

Table 1 lists evaluation results of 4 performance indexes according to the method of the present invention. It is known from data listed in Table 1. Evaluation result of the present invention has excellent accuracy, monotonicity and consistency, and thus is capable of predicting visual subjective perception of human beings to stereo images well.

TABLE 1 Evaluation result of the performance indexes according to the method of the present invention All Gblur JPEG JP2K Wn H.264 data Pearson linear correlation 0.965 0.949 0.928 0.968 0.964 0.938 coefficient (CC) Spearman rank correlation 0.952 0.950 0.934 0.941 0.957 0.939 coefficient (SROCC) root-mean-square error 5.209 4.388 4.169 3.764 3.230 5.606 (RMSE) obnormal value ratio (OR) 0 0 0 0 0 0

One skilled in the art will understand that the embodiment of the present invention as shown in the drawings and described above is exemplary only and not intended to be limiting.

It will thus be seen that the objects of the present invention have been fully and effectively accomplished. Its embodiments have been shown and described for the purposes of illustrating the functional and structural principles of the present invention and is subject to change without departure from such principles. Therefore, this invention includes all modifications encompassed within the spirit and scope of the following claims.

## Claims

1. A method for objectively evaluating quality of a stereo image comprising following steps of:
{circle around (1)} defining an original stereo image which is utilized for referring in an objective quality evaluation of the stereo image as a reference stereo image, a left-view of the reference stereo image is denoted as Il, a right-view of the reference stereo image is denoted as Ir a cyclopean image of the reference stereo image formed in a human visual system is denoted as C;
defining a distortion image of the reference stereo image which is used for testing in the objective quality evaluation of the stereo image as a testing stereo image, a left-view of the testing stereo image is denoted as Îl, a right-view of the testing stereo image is denoted as Îr, a cyclopean image of the testing stereo image formed in the human visual system is denoted as Ĉ, wherein sizes of the left-view of the reference stereo image Il, the right-view of the reference stereo image Ir, the left-view of the testing stereo image Îl, and the right-view of the testing stereo image Îr, are all denoted as W×H, wherein W refers to a width thereof, and H refers to a height thereof;
{circle around (2)} classifying the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three areas, wherein:
a first area thereof is an occlusion area, which is denoted as {circumflex over (R)}occ, information of the occlusion area {circumflex over (R)}occ exists only in the left-view or the right-view of the testing stereo image;
a second area thereof is a binocular suppression area, which is denoted as {circumflex over (R)}bs, information of the binocular suppression area {circumflex over (R)}bs is binocular information having great image content difference or great disparity between the left-view and the right-view of the testing stereo image; and
a third area thereof is a binocular fusion area, which is denoted as {circumflex over (R)}bf, information of the binocular fusion area {circumflex over (R)}bf is binocular information having similar image content and small disparity between the left-view and the right-view of the testing stereo image;
{circle around (3)} dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,
classifying all of the image blocks into three types: an occlusion block, a binocular suppression block and a binocular fusion block,
updating {circumflex over (R)}occ, {circumflex over (R)}bs and {circumflex over (R)}bf according to types of the image blocks, so as to obtain an updated occlusion area {circumflex over (R)}occ′, an updated binocular suppression area {circumflex over (R)}bs′ and an updated binocular fusion area {circumflex over (R)}bf′,
in the cyclopean image C of the reference stereo image formed in the human visual system, obtaining an area Rocc′ corresponding to a position of {circumflex over (R)}occ′, an area Rbs′ corresponding to a position of {circumflex over (R)}bs′, and an area Rbf′ corresponding to a position of {circumflex over (R)}bf′,
processing singular value decomposition respectively on all of the image blocks having a size of k×k in the area {circumflex over (R)}occ′, an area Rbsl′ corresponding to left-view of {circumflex over (R)}bs′, an area {circumflex over (R)}bsr′ corresponding to right-view of {circumflex over (R)}bs′, an area {circumflex over (R)}bfl′ corresponding to left-view of {circumflex over (R)}bf′, an area {circumflex over (R)}bfr′ corresponding to right-view of {circumflex over (R)}bf′, the area Rocc′, an area Rbsl′ corresponding to left-view of Rbs′, an area Rbsr′ corresponding to right-view of Rbs′, an area Rbfl′ corresponding to left-view of Rbf′, and an area Rbfr′ corresponding to right-view of Rbf′, in such a manner that singular values respectively corresponding to each image block are obtained, wherein k>1;
{circle around (4)} obtaining an overall distortion degree Qocc of {circumflex over (R)}occ′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}occ′ and {circumflex over (R)}occ′ and have a size of k×k,
obtaining an overall distortion degree Qbs of {circumflex over (R)}bs′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsl′ and Rb,sl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsr′ and Rbsr′ and have a size of k×k, and
obtaining an overall distortion degree Qbf of {circumflex over (R)}bf′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bfl′ and Rbfl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have a size of k×k; and
{circle around (5)} according to the overall distortion degree Qocc of {circumflex over (R)}occ′, the overall distortion degree Qbs of {circumflex over (R)}bs′ and the overall distortion degree Qbf of {circumflex over (R)}bf′, calculating an overall distortion degree Q of the testing stereo image by an equation of linear regression, wherein Q=a×Qocc+b×Qbs+c×Qbf, wherein a, b and c are all coefficients and satisfy:
${ a + b + c = 1 0 ≤ a ≤ 1 0 ≤ b ≤ 1 0 ≤ c ≤ 1.$
{circle around (1)} defining an original stereo image which is utilized for referring in an objective quality evaluation of the stereo image as a reference stereo image, a left-view of the reference stereo image is denoted as Il, a right-view of the reference stereo image is denoted as Ir a cyclopean image of the reference stereo image formed in a human visual system is denoted as C;
defining a distortion image of the reference stereo image which is used for testing in the objective quality evaluation of the stereo image as a testing stereo image, a left-view of the testing stereo image is denoted as Îl, a right-view of the testing stereo image is denoted as Îr, a cyclopean image of the testing stereo image formed in the human visual system is denoted as Ĉ, wherein sizes of the left-view of the reference stereo image Il, the right-view of the reference stereo image Ir, the left-view of the testing stereo image Îl, and the right-view of the testing stereo image Îr, are all denoted as W×H, wherein W refers to a width thereof, and H refers to a height thereof;
{circle around (2)} classifying the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three areas, wherein:
a first area thereof is an occlusion area, which is denoted as {circumflex over (R)}occ, information of the occlusion area {circumflex over (R)}occ exists only in the left-view or the right-view of the testing stereo image;
a second area thereof is a binocular suppression area, which is denoted as {circumflex over (R)}bs, information of the binocular suppression area {circumflex over (R)}bs is binocular information having great image content difference or great disparity between the left-view and the right-view of the testing stereo image; and
a third area thereof is a binocular fusion area, which is denoted as {circumflex over (R)}bf, information of the binocular fusion area {circumflex over (R)}bf is binocular information having similar image content and small disparity between the left-view and the right-view of the testing stereo image;
{circle around (3)} dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,
classifying all of the image blocks into three types: an occlusion block, a binocular suppression block and a binocular fusion block,
updating {circumflex over (R)}occ, {circumflex over (R)}bs and {circumflex over (R)}bf according to types of the image blocks, so as to obtain an updated occlusion area {circumflex over (R)}occ′, an updated binocular suppression area {circumflex over (R)}bs′ and an updated binocular fusion area {circumflex over (R)}bf′,
in the cyclopean image C of the reference stereo image formed in the human visual system, obtaining an area Rocc′ corresponding to a position of {circumflex over (R)}occ′, an area Rbs′ corresponding to a position of {circumflex over (R)}bs′, and an area Rbf′ corresponding to a position of {circumflex over (R)}bf′,
processing singular value decomposition respectively on all of the image blocks having a size of k×k in the area {circumflex over (R)}occ′, an area Rbsl′ corresponding to left-view of {circumflex over (R)}bs′, an area {circumflex over (R)}bsr′ corresponding to right-view of {circumflex over (R)}bs′, an area {circumflex over (R)}bfl′ corresponding to left-view of {circumflex over (R)}bf′, an area {circumflex over (R)}bfr′ corresponding to right-view of {circumflex over (R)}bf′, the area Rocc′, an area Rbsl′ corresponding to left-view of Rbs′, an area Rbsr′ corresponding to right-view of Rbs′, an area Rbfl′ corresponding to left-view of Rbf′, and an area Rbfr′ corresponding to right-view of Rbf′, in such a manner that singular values respectively corresponding to each image block are obtained, wherein k>1;
{circle around (4)} obtaining an overall distortion degree Qocc of {circumflex over (R)}occ′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}occ′ and {circumflex over (R)}occ′ and have a size of k×k,
obtaining an overall distortion degree Qbs of {circumflex over (R)}bs′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsl′ and Rb,sl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bsr′ and Rbsr′ and have a size of k×k, and
obtaining an overall distortion degree Qbf of {circumflex over (R)}bf′ according to singular value distances between all of the image blocks which are in corresponding positions in {circumflex over (R)}bfl′ and Rbfl′ and have a size of k×k, and singular value distances between all of the image blocks which are in corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have a size of k×k; and
{circle around (5)} according to the overall distortion degree Qocc of {circumflex over (R)}occ′, the overall distortion degree Qbs of {circumflex over (R)}bs′ and the overall distortion degree Qbf of {circumflex over (R)}bf′, calculating an overall distortion degree Q of the testing stereo image by an equation of linear regression, wherein Q=a×Qocc+b×Qbs+c×Qbf, wherein a, b and c are all coefficients and satisfy:
2. The method for objectively evaluating quality of a stereo image, as recited in claim 1, wherein specific process of the step {circle around (2)} comprises:
{circle around (2)}-1 denoting a pixel having a position coordinate of (x,y) in the left-view image Il of the reference stereo image as px,yl, denoting a pixel having a position coordinate of (s,t) in the right-view image Ir of the reference stereo image as ps,tr, denoting a pixel having a position coordinate of (x,y) in the left-view image Îl of the testing stereo image as {circumflex over (p)}x,yl, and denoting a pixel having a position coordinate of (s,t) in the right-view image Îr of the testing stereo image as {circumflex over (p)}s,tr, wherein 1≦x≦W, 1≦y≦H, 1≦s≦W and 1≦t≦H;
{circle around (2)}-2 processing stereo matching on the reference stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the reference stereo image, wherein a specific process thereof comprises:
judging whether px,yl is matched with ps,tr, wherein:
if px,yl is matched with ps,tr, matching relationship of px,yl and ps,tr is presented as (px,yl,ps,tr), horizontal disparities of px,yl and ps,tr are both denoted as dx,yh, wherein dx,yh=|s−x|, vertical disparities of px,yl and ps,tr are both denoted as dx,yv, wherein dx,yv=|t−y|,
if no pixel is matched with px,yl in the right-view Ir of the reference stereo image, matching relationship of px,yl is presented as (px,yl,φ), and the horizontal disparity and the vertical disparity of px,yl are both determined to be 255, and
if no pixel is matched with ps,tr in the left-view image Il of the reference stereo image, matching relationship of ps,tr is presented as (φ,ps,tr), and the horizontal disparity and the vertical disparity of ps,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign;
{circle around (3)}-3 processing stereo matching on the testing stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the testing stereo image, wherein a specific process thereof comprising:
judging whether {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr, wherein:
if {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr, matching relationship of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr is presented as ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr), the horizontal disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yh, wherein {circumflex over (d)}x,yh=|s−x|, the vertical disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yv, wherein {circumflex over (d)}x,yv=|t−y|,
if no pixel is matched with {circumflex over (p)}x,yl in the right-view Îr of the testing stereo image, matching relationship of {circumflex over (p)}x,yl is presented as ({circumflex over (p)}x,yl,φ), and the horizontal disparity and the vertical disparity of {circumflex over (p)}x,yl are both determined to be 255, and
if no pixel is matched with {circumflex over (p)}s,tr in the left-view Îl of the testing stereo image, matching relationship of {circumflex over (p)}s,tr is presented as (φ,{circumflex over (p)}s,tr), and the horizontal disparity and the vertical disparity of {circumflex over (p)}s,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign;
{circle around (2)}-4 dividing the cyclopean image of the testing stereo image formed in the human visual system Ĉ into an occlusion {circumflex over (R)}occ and a match area {circumflex over (R)}match, wherein the occlusion area {circumflex over (R)}occ comprises a left-view occlusion area {circumflex over (R)}occl and a right-view occlusion area {circumflex over (R)}occr, wherein {circumflex over (R)}occ={circumflex over (R)}occl∪{circumflex over (R)}occr,
wherein the left-view occlusion area hoicc is a set of pixels having a matching relationship of ({circumflex over (p)}x,yl,φ) in the left-view Îl of the testing stereo image, which are presented as {circumflex over (R)}occl={{circumflex over (p)}x,yl|({circumflex over (p)}x,yl,φ){circumflex over (p)}x,ylεÎl},
the right-view occlusion area {circumflex over (R)}occr is a set of pixels having a matching relationship of (φ,{circumflex over (p)}s,tr) in the right-view Îr of the testing stereo image, which is presented as {circumflex over (R)}occr={{circumflex over (p)}s,tr|(φ, {circumflex over (p)}s,tr){circumflex over (p)}s,trεÎr},
the match area {circumflex over (R)}match is a set of matched pixels of the left-view Îl of the testing stereo image and the right-view Îr of the testing stereo image, which having a matching relationship of ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr), wherein “∪” is an operational sign of a union, “” is an operational sign of “and”; and
{circle around (2)}-5 further dividing the match area {circumflex over (R)}match into a binocular suppression area {circumflex over (R)}bs and a binocular fusion area {circumflex over (R)}bf, wherein:
the binocular suppression area {circumflex over (R)}bs is an integrated area formed by processing binocular suppression on the left-view corresponding area {circumflex over (R)}bsl of {circumflex over (R)}bs and the right-view corresponding area {circumflex over (R)}bsr of {circumflex over (R)}bs, and
the binocular fusion area {circumflex over (R)}bf is an integrated area formed by processing binocular fusion on the left-view corresponding area {circumflex over (R)}bfl of {circumflex over (R)}bf and the right-view corresponding view {circumflex over (R)}bfr of {circumflex over (R)}bf, wherein:

{circumflex over (R)}bsl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},

{circumflex over (R)}bsr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},

{circumflex over (R)}bfl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}, and

{circumflex over (R)}bfr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}.
{circle around (2)}-1 denoting a pixel having a position coordinate of (x,y) in the left-view image Il of the reference stereo image as px,yl, denoting a pixel having a position coordinate of (s,t) in the right-view image Ir of the reference stereo image as ps,tr, denoting a pixel having a position coordinate of (x,y) in the left-view image Îl of the testing stereo image as {circumflex over (p)}x,yl, and denoting a pixel having a position coordinate of (s,t) in the right-view image Îr of the testing stereo image as {circumflex over (p)}s,tr, wherein 1≦x≦W, 1≦y≦H, 1≦s≦W and 1≦t≦H;
{circle around (2)}-2 processing stereo matching on the reference stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the reference stereo image, wherein a specific process thereof comprises:
judging whether px,yl is matched with ps,tr, wherein:
if px,yl is matched with ps,tr, matching relationship of px,yl and ps,tr is presented as (px,yl,ps,tr), horizontal disparities of px,yl and ps,tr are both denoted as dx,yh, wherein dx,yh=|s−x|, vertical disparities of px,yl and ps,tr are both denoted as dx,yv, wherein dx,yv=|t−y|,
if no pixel is matched with px,yl in the right-view Ir of the reference stereo image, matching relationship of px,yl is presented as (px,yl,φ), and the horizontal disparity and the vertical disparity of px,yl are both determined to be 255, and
if no pixel is matched with ps,tr in the left-view image Il of the reference stereo image, matching relationship of ps,tr is presented as (φ,ps,tr), and the horizontal disparity and the vertical disparity of ps,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign;
{circle around (3)}-3 processing stereo matching on the testing stereo image, so as to obtain a horizontal disparity and a vertical disparity of each pixel of the testing stereo image, wherein a specific process thereof comprising:
judging whether {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr, wherein:
if {circumflex over (p)}x,yl is matched with {circumflex over (p)}s,tr, matching relationship of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr is presented as ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr), the horizontal disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yh, wherein {circumflex over (d)}x,yh=|s−x|, the vertical disparities of {circumflex over (p)}x,yl and {circumflex over (p)}s,tr are both denoted as {circumflex over (d)}x,yv, wherein {circumflex over (d)}x,yv=|t−y|,
if no pixel is matched with {circumflex over (p)}x,yl in the right-view Îr of the testing stereo image, matching relationship of {circumflex over (p)}x,yl is presented as ({circumflex over (p)}x,yl,φ), and the horizontal disparity and the vertical disparity of {circumflex over (p)}x,yl are both determined to be 255, and
if no pixel is matched with {circumflex over (p)}s,tr in the left-view Îl of the testing stereo image, matching relationship of {circumflex over (p)}s,tr is presented as (φ,{circumflex over (p)}s,tr), and the horizontal disparity and the vertical disparity of {circumflex over (p)}s,tr are both determined to be 255, wherein “∥” is an absolute value sign, “φ” is an empty set sign;
{circle around (2)}-4 dividing the cyclopean image of the testing stereo image formed in the human visual system Ĉ into an occlusion {circumflex over (R)}occ and a match area {circumflex over (R)}match, wherein the occlusion area {circumflex over (R)}occ comprises a left-view occlusion area {circumflex over (R)}occl and a right-view occlusion area {circumflex over (R)}occr, wherein {circumflex over (R)}occ={circumflex over (R)}occl∪{circumflex over (R)}occr,
wherein the left-view occlusion area hoicc is a set of pixels having a matching relationship of ({circumflex over (p)}x,yl,φ) in the left-view Îl of the testing stereo image, which are presented as {circumflex over (R)}occl={{circumflex over (p)}x,yl|({circumflex over (p)}x,yl,φ){circumflex over (p)}x,ylεÎl},
the right-view occlusion area {circumflex over (R)}occr is a set of pixels having a matching relationship of (φ,{circumflex over (p)}s,tr) in the right-view Îr of the testing stereo image, which is presented as {circumflex over (R)}occr={{circumflex over (p)}s,tr|(φ, {circumflex over (p)}s,tr){circumflex over (p)}s,trεÎr},
the match area {circumflex over (R)}match is a set of matched pixels of the left-view Îl of the testing stereo image and the right-view Îr of the testing stereo image, which having a matching relationship of ({circumflex over (p)}x,yl,{circumflex over (p)}s,tr), wherein “∪” is an operational sign of a union, “” is an operational sign of “and”; and
{circle around (2)}-5 further dividing the match area {circumflex over (R)}match into a binocular suppression area {circumflex over (R)}bs and a binocular fusion area {circumflex over (R)}bf, wherein:
the binocular suppression area {circumflex over (R)}bs is an integrated area formed by processing binocular suppression on the left-view corresponding area {circumflex over (R)}bsl of {circumflex over (R)}bs and the right-view corresponding area {circumflex over (R)}bsr of {circumflex over (R)}bs, and
the binocular fusion area {circumflex over (R)}bf is an integrated area formed by processing binocular fusion on the left-view corresponding area {circumflex over (R)}bfl of {circumflex over (R)}bf and the right-view corresponding view {circumflex over (R)}bfr of {circumflex over (R)}bf, wherein:

{circumflex over (R)}bsl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},

{circumflex over (R)}bsr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},

{circumflex over (R)}bfl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}, and

{circumflex over (R)}bfr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}.

{circumflex over (R)}bsl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},

{circumflex over (R)}bsr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|>|dx,yh|+|dx,yv|},

{circumflex over (R)}bfl={{circumflex over (p)}x,yl|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}, and

{circumflex over (R)}bfr={{circumflex over (p)}s,tr|{circumflex over (p)}x,ylεÎl{circumflex over (p)}s,trεÎr({circumflex over (p)}x,yl,{circumflex over (p)}s,tr)|{circumflex over (d)}x,yh|+|{circumflex over (d)}x,yv|≦|dx,yh|+|dx,yv|}.
3. The method for objectively evaluating quality of a stereo image, as recited in claim 1, wherein specific process of the step {circle around (3)} comprises:
{circle around (3)}-1 dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,
classifying all of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three types: an occlusion block, a binocular suppression block and a binocular fusion block, wherein specific process of the classifying comprises:
denoting any one of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as {circumflex over (B)}, and
judging whether there is a pixel belonging to {circumflex over (R)}occ in the {circumflex over (B)}, wherein if yes, mark {circumflex over (B)} as a occlusion block; if no, further judge whether there is a pixel belonging to {circumflex over (R)}bs in {circumflex over (B)}, if yes, mark {circumflex over (B)} as a binocular suppression block, if no mark {circumflex over (B)} as a binocular fusion block, wherein k>1;
{circle around (3)}-2 defining an area constituted by all of the occlusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated occlusion area, which is denoted as {circumflex over (R)}occ′,
defining an area constituted by all of the binocular suppression blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular suppression area, which is denoted as {circumflex over (R)}bf′, and
defining an area constituted by all of the binocular fusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular fusion area, which is denoted as {circumflex over (R)}bf′,
wherein the left-view corresponding area of the updated occlusion area {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occl′, the right-view corresponding area of {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occr′, left-view corresponding area of the updated binocular suppression area {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the right-view corresponding area of {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the left-view corresponding area of the updated binocular fusion area {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfl′, and the right-view corresponding area of {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfr′;
{circle around (3)}-3 denoting the area corresponding to the position of {circumflex over (R)}occ′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rocc′, wherein Rocc′=Roccl′∪Roccr′, wherein “∪” is an operational sign for union in a set,
wherein Roccl′ is the area corresponding to the position of Roccl′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}occl′},
wherein Roccr′ is the area corresponding to the position of {circumflex over (R)}occr′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}occr′};
{circle around (3)}-4 denoting the area corresponding to the position of {circumflex over (R)}bsl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsl′, wherein Rbsl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bsl′}, and
denoting the area corresponding to the position of {circumflex over (R)}b,sr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsr′, wherein Rbsr′={ps,tr|ps,trlεIr{circumflex over (p)}s,trε{circumflex over (R)}bsr′};
{circle around (3)}-5 denoting the area corresponding to the position of {circumflex over (R)}bfl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfl′, wherein Rbfl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bfl′}, and
denoting the area corresponding to the position of {circumflex over (R)}bfr′ in the cyclopean image C of the reference stereo image formed in the human visual system as {circumflex over (R)}bfr′, wherein Rbfr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bfr′}; and
{circle around (3)}-6 processing the singular value decomposition respectively on all of the image blocks having the size of k×k in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′, in such a manner that the singular values respectively corresponding to all image blocks having the size of k×k and in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ are obtained.
{circle around (3)}-1 dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,
classifying all of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three types: an occlusion block, a binocular suppression block and a binocular fusion block, wherein specific process of the classifying comprises:
denoting any one of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as {circumflex over (B)}, and
judging whether there is a pixel belonging to {circumflex over (R)}occ in the {circumflex over (B)}, wherein if yes, mark {circumflex over (B)} as a occlusion block; if no, further judge whether there is a pixel belonging to {circumflex over (R)}bs in {circumflex over (B)}, if yes, mark {circumflex over (B)} as a binocular suppression block, if no mark {circumflex over (B)} as a binocular fusion block, wherein k>1;
{circle around (3)}-2 defining an area constituted by all of the occlusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated occlusion area, which is denoted as {circumflex over (R)}occ′,
defining an area constituted by all of the binocular suppression blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular suppression area, which is denoted as {circumflex over (R)}bf′, and
defining an area constituted by all of the binocular fusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular fusion area, which is denoted as {circumflex over (R)}bf′,
wherein the left-view corresponding area of the updated occlusion area {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occl′, the right-view corresponding area of {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occr′, left-view corresponding area of the updated binocular suppression area {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the right-view corresponding area of {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the left-view corresponding area of the updated binocular fusion area {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfl′, and the right-view corresponding area of {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfr′;
{circle around (3)}-3 denoting the area corresponding to the position of {circumflex over (R)}occ′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rocc′, wherein Rocc′=Roccl′∪Roccr′, wherein “∪” is an operational sign for union in a set,
wherein Roccl′ is the area corresponding to the position of Roccl′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}occl′},
wherein Roccr′ is the area corresponding to the position of {circumflex over (R)}occr′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}occr′};
{circle around (3)}-4 denoting the area corresponding to the position of {circumflex over (R)}bsl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsl′, wherein Rbsl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bsl′}, and
denoting the area corresponding to the position of {circumflex over (R)}b,sr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsr′, wherein Rbsr′={ps,tr|ps,trlεIr{circumflex over (p)}s,trε{circumflex over (R)}bsr′};
{circle around (3)}-5 denoting the area corresponding to the position of {circumflex over (R)}bfl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfl′, wherein Rbfl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bfl′}, and
denoting the area corresponding to the position of {circumflex over (R)}bfr′ in the cyclopean image C of the reference stereo image formed in the human visual system as {circumflex over (R)}bfr′, wherein Rbfr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bfr′}; and
{circle around (3)}-6 processing the singular value decomposition respectively on all of the image blocks having the size of k×k in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′, in such a manner that the singular values respectively corresponding to all image blocks having the size of k×k and in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ are obtained.
4. The method for objectively evaluating quality of a stereo image, as recited in claim 2, wherein specific process of the step {circle around (3)} comprises:
{circle around (3)}-1 dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,
classifying all of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three types: an occlusion block, a binocular suppression block and a binocular fusion block, wherein specific process of the classifying comprises:
denoting any one of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as {circumflex over (B)}, and
judging whether there is a pixel belonging to {circumflex over (R)}occ in the {circumflex over (B)}, wherein if yes, mark {circumflex over (B)} as a occlusion block; if no, further judge whether there is a pixel belonging to {circumflex over (R)}bs in {circumflex over (B)}, if yes, mark {circumflex over (B)} as a binocular suppression block, if no mark {circumflex over (B)} as a binocular fusion block, wherein k>1;
{circle around (3)}-2 defining an area constituted by all of the occlusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated occlusion area, which is denoted as {circumflex over (R)}occ′,
defining an area constituted by all of the binocular suppression blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular suppression area, which is denoted as {circumflex over (R)}bf′, and
defining an area constituted by all of the binocular fusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular fusion area, which is denoted as {circumflex over (R)}bf′,
wherein the left-view corresponding area of the updated occlusion area {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occl′, the right-view corresponding area of {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occr′ the left-view corresponding area of the updated binocular suppression area {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the right-view corresponding area of {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsr′, the left-view corresponding area of the updated binocular fusion area {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfl′, and the right-view corresponding area of {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfr′;
{circle around (3)}-3 denoting the area corresponding to the position of {circumflex over (R)}occ′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rocc′, wherein Rocc′=Roccl′∪Roccr′, wherein “∪” is an operational sign for union in a set,
wherein Roccl′ is the area corresponding to the position of {circumflex over (R)}occl′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}occl′},
wherein Roccr′ is the area corresponding to the position of {circumflex over (R)}occr′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}occr′};
{circle around (3)}-4 denoting the area corresponding to the position of {circumflex over (R)}bsl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsl′, wherein Rbsl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bsl′}, and
denoting the area corresponding to the position of {circumflex over (R)}bsr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsr′, wherein Rbsr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bsr′};
{circle around (3)}-5 denoting the area corresponding to the position of {circumflex over (R)}bfl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfl′, wherein Rbfl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bfl′}, and
denoting the area corresponding to the position of {circumflex over (R)}bfr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfr′, wherein Rbfr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bfr′}; and
{circle around (3)}-6 processing the singular value decomposition respectively on all of the image blocks having the size of k×k in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ in such a manner that the singular values respectively corresponding to all image blocks having the size of k×k and in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ are obtained.
{circle around (3)}-1 dividing the cyclopean image Ĉ of the testing stereo image formed in the human visual system into a plurality of image blocks which are non-overlapped with each other and have a size of k×k,
classifying all of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system into three types: an occlusion block, a binocular suppression block and a binocular fusion block, wherein specific process of the classifying comprises:
denoting any one of the image blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as {circumflex over (B)}, and
judging whether there is a pixel belonging to {circumflex over (R)}occ in the {circumflex over (B)}, wherein if yes, mark {circumflex over (B)} as a occlusion block; if no, further judge whether there is a pixel belonging to {circumflex over (R)}bs in {circumflex over (B)}, if yes, mark {circumflex over (B)} as a binocular suppression block, if no mark {circumflex over (B)} as a binocular fusion block, wherein k>1;
{circle around (3)}-2 defining an area constituted by all of the occlusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated occlusion area, which is denoted as {circumflex over (R)}occ′,
defining an area constituted by all of the binocular suppression blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular suppression area, which is denoted as {circumflex over (R)}bf′, and
defining an area constituted by all of the binocular fusion blocks of the cyclopean image Ĉ of the testing stereo image formed in the human visual system as an updated binocular fusion area, which is denoted as {circumflex over (R)}bf′,
wherein the left-view corresponding area of the updated occlusion area {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occl′, the right-view corresponding area of {circumflex over (R)}occ′ is denoted as {circumflex over (R)}occr′ the left-view corresponding area of the updated binocular suppression area {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsl′, the right-view corresponding area of {circumflex over (R)}bs′ is denoted as {circumflex over (R)}bsr′, the left-view corresponding area of the updated binocular fusion area {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfl′, and the right-view corresponding area of {circumflex over (R)}bf′ is denoted as {circumflex over (R)}bfr′;
{circle around (3)}-3 denoting the area corresponding to the position of {circumflex over (R)}occ′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rocc′, wherein Rocc′=Roccl′∪Roccr′, wherein “∪” is an operational sign for union in a set,
wherein Roccl′ is the area corresponding to the position of {circumflex over (R)}occl′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}occl′},
wherein Roccr′ is the area corresponding to the position of {circumflex over (R)}occr′ in the cyclopean image C of the reference stereo image formed in the human visual system, Roccl′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}occr′};
{circle around (3)}-4 denoting the area corresponding to the position of {circumflex over (R)}bsl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsl′, wherein Rbsl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bsl′}, and
denoting the area corresponding to the position of {circumflex over (R)}bsr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbsr′, wherein Rbsr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bsr′};
{circle around (3)}-5 denoting the area corresponding to the position of {circumflex over (R)}bfl′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfl′, wherein Rbfl′={px,yl|px,ylεIl{circumflex over (p)}x,ylε{circumflex over (R)}bfl′}, and
denoting the area corresponding to the position of {circumflex over (R)}bfr′ in the cyclopean image C of the reference stereo image formed in the human visual system as Rbfr′, wherein Rbfr′={ps,tr|ps,trεIr{circumflex over (p)}s,trε{circumflex over (R)}bfr′}; and
{circle around (3)}-6 processing the singular value decomposition respectively on all of the image blocks having the size of k×k in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ in such a manner that the singular values respectively corresponding to all image blocks having the size of k×k and in {circumflex over (R)}occ′, {circumflex over (R)}bsl′, {circumflex over (R)}bsr′, {circumflex over (R)}bfl′, {circumflex over (R)}bfr′, Rocc′, Rbsl′, Rbsr′, Rbfl′ and Rbfr′ are obtained.
5. The method for objectively evaluating quality of a stereo image, as recited in claim 3, wherein in the step {circle around (3)}-1, k=4.
6. The method for objectively evaluating quality of a stereo image, as recited in claim 4, wherein in the step {circle around (3)}-1, k=4.
7. The method for objectively evaluating quality of a stereo image, as recited in claim 3, wherein specific process of the step {circle around (4)} comprises:
{circle around (4)}-1 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}occ′ and Rocc′ and have the size of k×k, wherein an ith occlusion block in {circumflex over (R)}occ′ is denoted as {circumflex over (B)}i, in Rocc′ an ith occlusion block in a corresponding position of {circumflex over (B)}i is denoted as Bi, a singular value distance between {circumflex over (B)}i and Bi is calculated and is denoted as Docc(i),
$D occ ⁡ ( i ) = ∑ j = 1 k ⁢ ⁢ ( s i , j - s ^ i , j ) 2 ,$
calculating the singular value distances corresponding to all of the occlusion blocks in {circumflex over (R)}occ′ as the overall distortion degree of {circumflex over (R)}occ′, which is denoted as Qocc, wherein
$Q occ = 1 N occ ⁢ ∑ i = 1 N occ ⁢ ⁢  D occ ⁡ ( i ) - D occ ⁡ ( m )  ,$
wherein 1≦i≦Nocc, Nocc refers to a number of all of the occlusion blocks in {circumflex over (R)}occ′, and Nocc is also a number of all of the occlusion blocks in Rocc′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si,j refers to a j th singular value of Bi, ŝi,j refers to the jth singular value of {circumflex over (B)}i, “∥” is an absolute value sign, and Docc(m) refers to a median of Docc(1), Docc(2), . . . , Docc(Nocc);
{circle around (4)}-2 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsl′ and Rbsl′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsl′ is denoted as Bi′l in Rbsl′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′l is denoted as Bi′l, a singular value distance between {circumflex over (B)}i′l and Bi′l is calculated and is denoted as Dbsl(i′),
$D bs l ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ⁢ ( s i ′ , j l - s ^ i ′ , j l ) 2 ,$
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsr′ and Rbsr′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsr′ is denoted as {circumflex over (B)}i′r, in Rbsr′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′r is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r and Bi′r is calculated and is denoted as Dbsr(i′),
$D bs r ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ( s i ′ , j r - s ^ i ′ , j r ) 2 ,$
calculating the singular value distances of all of the binocular suppression blocks in {circumflex over (R)}bs′, wherein an i′th binocular suppression block in {circumflex over (R)}bs′ is denoted as {circumflex over (B)}i′, and its singular value distance is calculated and denoted as Dbs (i′), Dbs(i′)=min{Dbsl(i′), Dbsr(i′)}, and
calculating the singular value distances corresponding to all of the binocular suppression blocks in {circumflex over (R)}bs′ as the overall distortion degree of {circumflex over (R)}bs′, which is denoted as Qbs, wherein
$Q bs = 1 N bs ⁢ ∑ i ′ = 1 N bs ⁢  D bs ⁡ ( i ′ ) - D bs ⁡ ( m )  ,$
wherein 1≦i′≦Nbs, Nbs refers to a number of all of the binocular suppression blocks in {circumflex over (R)}bsl′ or {circumflex over (R)}bsr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si′,jl refers to a j th singular value of Bi′r, ŝi′,jr refers to the jth singular value of {circumflex over (B)}i′l, si′,jr refers to a j th singular value of Bi′r, ŝi′,jr refers to a jth singular value of {circumflex over (B)}i′r, min ( ) is a sign for a minimum value, “∥” an absolute value sign, and Dbs(m) refers to a median of Dbs(1), Dbs(2), . . . , Dbs(Nbs); and
{circle around (4)}-3 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfl′ and Rbfl′ and have the size of k×k, wherein an i″th binocular fusion block {circumflex over (R)}bfl′ is denoted as {circumflex over (B)}i″l′, in Rbfl′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″l is denoted as Bi″l, a singular value distance between {circumflex over (B)}i″l′ and Bi″l is calculated and is denoted as Dbfl(i″),
$D bf l ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ( s i ″ , j l - s ^ i ″ , j l ) 2 ,$
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfr′ is denoted as {circumflex over (B)}i″r′ in Rbfr′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″r′ is denoted as Bi″r, a singular value distance between {circumflex over (B)}i″r′ and Bi′r is calculated and is denoted as Dbfr(i″),
$D bf r ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ( s i ″ , j r - s ^ i ″ , j r ) 2 ,$
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfl′ for serving as an overall distortion degree of {circumflex over (R)}bfl′, which is denoted as Qbfl,
$Q bf l = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢  D bf l ⁡ ( i ″ ) - D bf l ⁡ ( m )  ,$
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfr′ for serving as an overall distortion degree of {circumflex over (R)}bfr′, which is denoted as Qbfr,
$Q bf r = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢  D bf r ⁡ ( i ″ ) - D bf r ⁡ ( m )  ,$
and
calculating an overall distortion degree of {circumflex over (R)}bf′ according to Qbfl and Qbfr, which is denoted as Qbf, Qbf=0.7×(Qbfl+Qbfr),
wherein 1≦i″≦Nbf, Nbf refers to a number of all of the binocular fusion blocks in {circumflex over (R)}bfl′ or {circumflex over (R)}bfr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si′,jl refers to a j th singular value of Bi″l, ŝi″,jl refers to the jth singular value of {circumflex over (B)}i″l′, si″,jr refers to a jth singular value of Bi″l, ŝi″,jl refers to a jth singular value of {circumflex over (B)}i″r, “∥” is an absolute value sign, and Dbfl(m) refers to a median of Dbfl(1), Dbfl(2), . . . , Dbfl(Nbf), Dbfr(m) refers to a median of Dbfr(1), Dbfr(2), . . . , Dbfr(Nbf).
{circle around (4)}-1 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}occ′ and Rocc′ and have the size of k×k, wherein an ith occlusion block in {circumflex over (R)}occ′ is denoted as {circumflex over (B)}i, in Rocc′ an ith occlusion block in a corresponding position of {circumflex over (B)}i is denoted as Bi, a singular value distance between {circumflex over (B)}i and Bi is calculated and is denoted as Docc(i),
calculating the singular value distances corresponding to all of the occlusion blocks in {circumflex over (R)}occ′ as the overall distortion degree of {circumflex over (R)}occ′, which is denoted as Qocc, wherein
{circle around (4)}-2 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsl′ and Rbsl′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsl′ is denoted as Bi′l in Rbsl′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′l is denoted as Bi′l, a singular value distance between {circumflex over (B)}i′l and Bi′l is calculated and is denoted as Dbsl(i′),
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsr′ and Rbsr′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsr′ is denoted as {circumflex over (B)}i′r, in Rbsr′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′r is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r and Bi′r is calculated and is denoted as Dbsr(i′),
calculating the singular value distances of all of the binocular suppression blocks in {circumflex over (R)}bs′, wherein an i′th binocular suppression block in {circumflex over (R)}bs′ is denoted as {circumflex over (B)}i′, and its singular value distance is calculated and denoted as Dbs (i′), Dbs(i′)=min{Dbsl(i′), Dbsr(i′)}, and
calculating the singular value distances corresponding to all of the binocular suppression blocks in {circumflex over (R)}bs′ as the overall distortion degree of {circumflex over (R)}bs′, which is denoted as Qbs, wherein
{circle around (4)}-3 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfl′ and Rbfl′ and have the size of k×k, wherein an i″th binocular fusion block {circumflex over (R)}bfl′ is denoted as {circumflex over (B)}i″l′, in Rbfl′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″l is denoted as Bi″l, a singular value distance between {circumflex over (B)}i″l′ and Bi″l is calculated and is denoted as Dbfl(i″),
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfr′ is denoted as {circumflex over (B)}i″r′ in Rbfr′ an i″th binocular fusion block in a corresponding position of {circumflex over (B)}i″r′ is denoted as Bi″r, a singular value distance between {circumflex over (B)}i″r′ and Bi′r is calculated and is denoted as Dbfr(i″),
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfl′ for serving as an overall distortion degree of {circumflex over (R)}bfl′, which is denoted as Qbfl,
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfr′ for serving as an overall distortion degree of {circumflex over (R)}bfr′, which is denoted as Qbfr,
calculating an overall distortion degree of {circumflex over (R)}bf′ according to Qbfl and Qbfr, which is denoted as Qbf, Qbf=0.7×(Qbfl+Qbfr),
wherein 1≦i″≦Nbf, Nbf refers to a number of all of the binocular fusion blocks in {circumflex over (R)}bfl′ or {circumflex over (R)}bfr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si′,jl refers to a j th singular value of Bi″l, ŝi″,jl refers to the jth singular value of {circumflex over (B)}i″l′, si″,jr refers to a jth singular value of Bi″l, ŝi″,jl refers to a jth singular value of {circumflex over (B)}i″r, “∥” is an absolute value sign, and Dbfl(m) refers to a median of Dbfl(1), Dbfl(2), . . . , Dbfl(Nbf), Dbfr(m) refers to a median of Dbfr(1), Dbfr(2), . . . , Dbfr(Nbf).
8. The method for objectively evaluating quality of a stereo image, as recited in claim 4, wherein specific process of the step {circle around (4)} comprises:
{circle around (4)}-1 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}occ′ and Rocc′ and have the size of k×k, wherein an ith occlusion block in {circumflex over (R)}occ′ is denoted as {circumflex over (B)}i, in Rocc′ an ith occlusion block in a corresponding position of {circumflex over (B)}i is denoted as Bi, a singular value distance between {circumflex over (B)}i and Bi is calculated and is denoted as Docc(i),
$D occ ⁡ ( i ) = ∑ j = 1 k ⁢ ( s i , j - s ^ i , j ) 2 ,$
calculating the singular value distances corresponding to all of the occlusion blocks in {circumflex over (R)}occ′ as the overall distortion degree of which is denoted as Qocc, wherein
$Q occ = 1 N occ ⁢ ∑ i = 1 N occ ⁢  D occ ⁡ ( i ) - D occ ⁡ ( m )  ,$
wherein 1≦i≦Nocc, Nocc refers to a number of all of the occlusion blocks in and {circumflex over (R)}occ′, and Nocc is also a number of all of the occlusion blocks in Rocc′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si,j refers to a j th singular value of Bi, ŝi,j refers to the jth singular value of {circumflex over (B)}i, “∥” is an absolute value sign, and Docc (m) refers to a median of Docc(1), Docc (2), . . . , Docc(Nocc);
{circle around (4)}-2 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsl′ and Rbsl′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsl′ is denoted as {circumflex over (B)}i′l, in Rbsl′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′l′ is denoted as Bi′l, a singular value distance between {circumflex over (B)}i′l and Bi′l is calculated and is denoted as Dbsl(i′),
$D bs l ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ( s i ′ , j l - s ^ i ′ , j l ) 2 ,$
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsr′ and Rbsr′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsr′ is denoted as {circumflex over (B)}i′r′, in Rbsr′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′r is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r and Bi′r is calculated and is denoted as Dbsr(i′),
$D bs r ⁡ ( i ′ ) = ∑ j = 1 k ⁢ ( s i ′ , j r - s ^ i ′ , j r ) 2 ,$
calculating the singular value distances of all of the binocular suppression blocks in {circumflex over (R)}bs′, wherein an i′th binocular suppression block in {circumflex over (R)}bs′ is denoted as {circumflex over (B)}i′, and its singular value distance is calculated and denoted as Dbs(i′), Dbs(i′)=min{Dbsl(i′), Dbsr(i′)}, and
calculating the singular value distances corresponding to all of the binocular suppression blocks in {circumflex over (R)}bs′ as the overall distortion degree of {circumflex over (R)}bs′, which is denoted as Qbs, wherein
$Q bs = 1 N bs ⁢ ∑ i ′ = 1 N bs ⁢  D bs ⁡ ( i ′ ) - D bs ⁡ ( m )  ,$
wherein 1≦i′≦Nbs, Nbs refers to a number of all of the binocular suppression blocks in {circumflex over (R)}bsl′ or {circumflex over (R)}bsr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si′,jl refers to a j th singular value of Bi′l, ŝi′,jl refers to the jth singular value of {circumflex over (B)}i′l, si′,jr refers to a j th singular value of Bi′r, ŝi′,jr refers to a jth singular value of {circumflex over (B)}i′l, min ( ) is a sign for a minimum value, “∥” is an absolute value sign, and Dbs(m) refers to a median of Dbs(1), Dbs(2), . . . , Dbs(Nbs); and
{circle around (4)}-3 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfl′ and Rbfl′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfl′ is denoted as {circumflex over (R)}i″l′ an i″th binocular fusion block in a corresponding position of {circumflex over (R)}i″l is denoted as Bi″l, a singular value distance between {circumflex over (R)}i″l and Bi″l is calculated and is denoted as Dbfl(i″)
$D bf l ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ( s i ″ , j l - s ^ i ″ , j l ) 2 ,$
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfr′ is denoted as {circumflex over (R)}i″r, an i″th binocular fusion block in a corresponding position of {circumflex over (R)}i″r′ is denoted as Bi″r, a singular value distance between {circumflex over (B)}i″r and Bi″r is calculated and is denoted as Dbfr(i″),
$D bf r ⁡ ( i ″ ) = ∑ j = 1 k ⁢ ( s i ″ , j r - s ^ i ″ , j r ) 2 ,$
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfl′ for serving as an overall distortion degree of {circumflex over (R)}bfl′, which is denoted as Qbfl,
$Q bf l = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢  D bf l ⁡ ( i ″ ) - D bf l ⁡ ( m )  ,$
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfr′ for serving as an overall distortion degree of {circumflex over (R)}bfr′, which is denoted as Qbfr,
$Q bf r = 1 N bf ⁢ ∑ i ″ = 1 N bf ⁢  D bf r ⁡ ( i ″ ) - D bf r ⁡ ( m )  ,$
and
calculating an overall distortion degree of {circumflex over (R)}bf′ according to Qbfl and Qbfr, which is denoted as Qbf, Qbf=0.7×(Qbfl+Qbfr),
wherein 1≦i″≦Nbf, Nbf refers to a number of all of the binocular fusion blocks in {circumflex over (R)}bfl′ or {circumflex over (R)}bfr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si″,jr refers to a j th singular value of Bi″l, ŝi″,jl refers to the jth singular value of {circumflex over (R)}i″l, si″,jr refers to a jth singular value of Bi″r, ŝi″,jr refers to a jth singular value of {circumflex over (B)}i″r, “∥” is an absolute value sign, and Dbfl(m) refers to a median of Dbfl(1), Dbfl(2), . . . , Dbfl(Nbf), Dbfr(m) refers to a median of Dbfr(1), Dbfr(2), . . . , Dbfr(Nbf).
{circle around (4)}-1 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}occ′ and Rocc′ and have the size of k×k, wherein an ith occlusion block in {circumflex over (R)}occ′ is denoted as {circumflex over (B)}i, in Rocc′ an ith occlusion block in a corresponding position of {circumflex over (B)}i is denoted as Bi, a singular value distance between {circumflex over (B)}i and Bi is calculated and is denoted as Docc(i),
calculating the singular value distances corresponding to all of the occlusion blocks in {circumflex over (R)}occ′ as the overall distortion degree of which is denoted as Qocc, wherein
{circle around (4)}-2 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsl′ and Rbsl′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsl′ is denoted as {circumflex over (B)}i′l, in Rbsl′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′l′ is denoted as Bi′l, a singular value distance between {circumflex over (B)}i′l and Bi′l is calculated and is denoted as Dbsl(i′),
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bsr′ and Rbsr′ and have the size of k×k, wherein an i′th binocular suppression block in {circumflex over (R)}bsr′ is denoted as {circumflex over (B)}i′r′, in Rbsr′ an i′th binocular suppression block in a corresponding position of {circumflex over (B)}i′r is denoted as Bi′r, a singular value distance between {circumflex over (B)}i′r and Bi′r is calculated and is denoted as Dbsr(i′),
calculating the singular value distances of all of the binocular suppression blocks in {circumflex over (R)}bs′, wherein an i′th binocular suppression block in {circumflex over (R)}bs′ is denoted as {circumflex over (B)}i′, and its singular value distance is calculated and denoted as Dbs(i′), Dbs(i′)=min{Dbsl(i′), Dbsr(i′)}, and
calculating the singular value distances corresponding to all of the binocular suppression blocks in {circumflex over (R)}bs′ as the overall distortion degree of {circumflex over (R)}bs′, which is denoted as Qbs, wherein
{circle around (4)}-3 calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfl′ and Rbfl′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfl′ is denoted as {circumflex over (R)}i″l′ an i″th binocular fusion block in a corresponding position of {circumflex over (R)}i″l is denoted as Bi″l, a singular value distance between {circumflex over (R)}i″l and Bi″l is calculated and is denoted as Dbfl(i″)
calculating the singular value distances between the image blocks which are in all corresponding positions of {circumflex over (R)}bfr′ and Rbfr′ and have the size of k×k, wherein an i″th binocular fusion block in {circumflex over (R)}bfr′ is denoted as {circumflex over (R)}i″r, an i″th binocular fusion block in a corresponding position of {circumflex over (R)}i″r′ is denoted as Bi″r, a singular value distance between {circumflex over (B)}i″r and Bi″r is calculated and is denoted as Dbfr(i″),
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfl′ for serving as an overall distortion degree of {circumflex over (R)}bfl′, which is denoted as Qbfl,
calculating the singular value distances of all of the binocular fusion blocks in {circumflex over (R)}bfr′ for serving as an overall distortion degree of {circumflex over (R)}bfr′, which is denoted as Qbfr,
calculating an overall distortion degree of {circumflex over (R)}bf′ according to Qbfl and Qbfr, which is denoted as Qbf, Qbf=0.7×(Qbfl+Qbfr),
wherein 1≦i″≦Nbf, Nbf refers to a number of all of the binocular fusion blocks in {circumflex over (R)}bfl′ or {circumflex over (R)}bfr′, 1≦j≦k, k refers to a number of singular values in one of the image blocks, si″,jr refers to a j th singular value of Bi″l, ŝi″,jl refers to the jth singular value of {circumflex over (R)}i″l, si″,jr refers to a jth singular value of Bi″r, ŝi″,jr refers to a jth singular value of {circumflex over (B)}i″r, “∥” is an absolute value sign, and Dbfl(m) refers to a median of Dbfl(1), Dbfl(2), . . . , Dbfl(Nbf), Dbfr(m) refers to a median of Dbfr(1), Dbfr(2), . . . , Dbfr(Nbf).
9. The method for objectively evaluating quality of the stereo image, as recited in claim 7, wherein in the step {circle around (5)} a=0, b=0.5, c=0.5.
10. The method for objectively evaluating quality of the stereo image, as recited in claim 8, wherein in the step {circle around (5)} a=0, b=0.5, c=0.5.