CN104994375A - Three-dimensional image quality objective evaluation method based on three-dimensional visual saliency - Google Patents

Abstract

The invention discloses a three-dimensional image quality objective evaluation method based on three-dimensional visual saliency. The three-dimensional image quality objective evaluation method comprises the following steps: optimizing a three-dimensional visual saliency map by simulating central offset and central concave characteristic, and extracting an optimized three-dimensional visual saliency map; acquiring a three-dimensional image comprehensive quality map according to a distorted three-dimensional image quality map and a quality map of an absolute difference map; and performing weighted summation on the three-dimensional image comprehensive quality map through the three-dimensional visual saliency map to obtain a distorted three-dimensional image quality objective evaluation value. The overall performance of the method on the aspect of three-dimensional image quality objective evaluation is superior to an algorithm in the prior art; practical feasibility is realized; a high three-dimensional image quality objective evaluation index is obtained; and a plurality of demands in practical application are met.

Description

Stereoscopic image quality objective evaluation method based on stereoscopic vision significance

Technical Field

The invention relates to the field of stereoscopic images, in particular to a stereoscopic image quality objective evaluation method based on stereoscopic vision significance.

Background

In recent years, the application of stereo image processing technology in the entertainment industry and the scientific field is more and more extensive, and stereo images provide more real stereo vision feeling for people, which is different from the visual feeling of plane images brought to people. However, the stereoscopic images can generate various distortions in the processes of acquisition, compression, storage, transportation, display and the like, and the stereoscopic vision feeling of people is affected. The quality of the stereo image, which is an inherent property of the stereo image, is an important index for judging the distortion of the stereo image. Although the subjective quality evaluation method has high accuracy, the method has the defects of time and labor waste and high cost, so that the establishment of the objective evaluation algorithm for the quality of the three-dimensional image, which can accurately and efficiently simulate the subjective evaluation result of human eyes, has important significance.

After years of research, researchers at home and abroad have proposed a plurality of objective evaluation algorithms for the quality of stereo images. First, some classical plane image quality evaluation algorithms such as Mean Squared Error (MSE), Peak Signal to Noise Ratio (PSNR), Structural Similarity (SSIM) [1] are directly applied to left and right viewpoints of a stereoscopic image, and a stereoscopic image quality objective evaluation value [2] is obtained by weighting left and right image qualities. Documents [3] and [4] perform stereoscopic image quality evaluation by fusing disparity information or depth information of a stereoscopic image with a 2D evaluation method in consideration of the influence of depth perception on stereoscopic image quality evaluation.

Currently, researchers have modeled some human visual characteristics that are easy to model, such as: the method is characterized by comprising the following steps of obtaining a three-dimensional image quality objective evaluation algorithm, and obtaining a brightness amplitude nonlinear characteristic, a multichannel characteristic, a contrast sensitivity characteristic, a masking effect and the like. With the gradual deepening of the knowledge of human visual systems in various research fields, the integration of more complex and advanced human visual characteristics into the objective evaluation model becomes a necessary development direction. Visual saliency, a high-level characteristic of the human visual system, refers to the difference in the intensity of attention that human eyes assign to different regions of an image. Since the salient region is a region that is relatively easy for the human eye to pay attention to, when a distorted stereoscopic image is viewed, distortion occurring in the salient region has a greater influence on the visual perception of the human eye. The visual saliency map provides an important basis for objective evaluation of the quality of the stereo image, and documents [5-8] apply the influence of the visual saliency to an objective evaluation algorithm of the quality of the stereo image. Document [6] performs quality evaluation on the drawn image in combination with the stereo saliency and the SSIM algorithm [1 ]. Document [7] performs empirical value evaluation of stereoscopic image quality by using a phase consistency map and a saliency map as features of a distorted stereoscopic image, and weighting a difference between the two feature maps using a local matching function to calculate a correlation between left and right viewpoints. Document [8] uses an SSIM algorithm to obtain a structural similarity map of a composite image of an original stereo image and a distorted stereo image, and the structural similarity map and a stereo saliency map are fused to obtain a stereo image quality evaluation index. Document [9] combines the left and right views of the original and distorted stereo images with their saliency maps to determine selective saliency maps of the left and right views of the original and distorted stereo images; then, adopting an SSIM algorithm to obtain a salient structure similarity map for the selective salient map, and obtaining an objective evaluation value of a single view by distributing different weights to an edge region, a smooth region and a texture region of the selective salient map; and finally, weighting and averaging the objective evaluation values of the left view and the right view to obtain the objective evaluation value of the quality of the stereo image. Document [5] [7] uses the disparity map of the original stereoscopic image when extracting the stereoscopic saliency map, ignoring the influence of the depth information distortion of the distorted stereoscopic image on the stereoscopic saliency map detection of the distorted stereoscopic image.

However, the above method does not consider the influence of other visual characteristics of human eyes and depth information on stereo image quality evaluation, so that the objective evaluation value cannot effectively reflect the subjective evaluation result.

Disclosure of Invention

The invention provides a stereoscopic image quality objective evaluation method based on stereoscopic vision significance, which accurately and effectively establishes an objective evaluation model reflecting the quality of a stereoscopic image of a subjective evaluation result according to a fused stereoscopic vision significance map and a stereoscopic image comprehensive quality map, and is described in detail as follows:

a stereoscopic image quality objective evaluation method based on stereoscopic visual saliency comprises the following steps:

optimizing the stereoscopic vision saliency map by simulating central deviation and fovea characteristics, and extracting the optimized stereoscopic vision saliency map;

obtaining a comprehensive quality image of the stereo image according to the quality images of the distorted stereo image and the absolute difference image;

and carrying out weighted summation on the stereo image comprehensive quality image through the stereo salient image to obtain an objective evaluation value of the quality of the distorted stereo image.

Before the step of optimizing the stereoscopic vision saliency map by simulating the central offset and the fovea characteristic and extracting the optimized stereoscopic vision saliency map, the objective evaluation method for the quality of the stereoscopic image further comprises the following steps:

and acquiring a stereoscopic vision saliency map of the distorted stereoscopic image by simulating the visual saliency.

The step of acquiring the stereoscopic vision saliency map of the distorted stereoscopic image by simulating the visual saliency specifically comprises the following steps:

converting the right view of the distorted stereo image from an RGB color space to a YCbCr color space, dividing the distorted stereo image into non-overlapping image blocks, and performing discrete cosine transform on the image blocks;

and normalizing the brightness contrast characteristic saliency map, the chroma contrast characteristic saliency map, the texture contrast characteristic saliency map and the absolute difference map of the distorted stereo image, and linearly combining the normalized maps to obtain the stereo visual saliency map of the distorted stereo image.

The stereo visual saliency map is optimized by simulating central deviation and fovea characteristics, and the step of extracting the optimized stereo visual saliency map specifically comprises the following steps:

simulating a central offset factor, and performing primary optimization on the stereoscopic vision saliency map;

and simulating the foveal characteristic, and finally optimizing the preliminarily optimized stereoscopic vision saliency map.

The step of obtaining the stereo image comprehensive quality map according to the quality maps of the distorted stereo image quality map and the absolute difference map specifically comprises the following steps:

extracting the phase similarity of the right view of the distorted stereo image;

extracting gradient similarity of a right view of the distorted stereo image;

respectively extracting the similarity of the chromaticity of the right view of the distorted stereo image;

calculating the similarity between the original stereo image right view and the distorted stereo image right view;

calculating the similarity of a left view of the distorted stereo image, and obtaining a quality image of the distorted stereo image by using the similarity of a right view of the distorted stereo image and a similarity image of the left view by using an average value method;

extracting a quality diagram of the absolute difference diagram of the distorted stereo image by adopting a structural similarity algorithm for the absolute difference diagram of the original stereo image and the absolute difference diagram of the distorted stereo image;

and fusing the quality image of the distorted stereo image with the quality image of the absolute difference image to obtain a comprehensive quality image of the stereo image.

The technical scheme provided by the invention has the beneficial effects that: the method obtains the optimal stereoscopic visual saliency map by adopting the characteristics of central deviation and fovea on the basis of the initial stereoscopic saliency map, and better simulates the visual process of a human visual system for extracting a salient region. And then, weighting and multiplying the stereoscopic vision saliency map and the stereoscopic image quality map, and normalizing to obtain the objective evaluation value of the stereoscopic image. Compared with the other three algorithms, the algorithm avoids the defect that the same weight is adopted for different areas of the whole image. Experimental verification shows that the overall performance of the method is superior to that of a 3SIQA [8] algorithm, a DSSIM3 algorithm and an SSIM [1] algorithm in the aspect of objective evaluation of the quality of the stereo image, and the method has practical feasibility.

Drawings

FIG. 1 is a schematic representation of a source image "tree 2";

FIG. 2 is a schematic diagram of a source image "girl";

FIG. 3 is a schematic representation of a source image "tree 1";

FIG. 4 is a schematic view of a source image "tju";

FIG. 5 is a schematic diagram of a source image "family";

FIG. 6 is a schematic view of a degraded image "river";

FIG. 7 is a schematic diagram of a source image "ox";

FIG. 8 is a schematic diagram of a source image "wman";

FIG. 9 is a block diagram of a stereoscopic image quality evaluation method based on stereoscopic saliency;

FIG. 10 is a comparison of an original saliency map, a saliency map combining two influencing factors;

(a) is a distorted stereoscopic image right view; (b) to the perspective saliency map SM1 before optimization; (c) is a three-dimensional saliency map SM' after central offset optimization; (d) the three-dimensional saliency map SM after the optimization of SM' through the fovea is obtained.

Fig. 11 is a flowchart of a stereoscopic image quality objective evaluation method based on stereoscopic saliency.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, embodiments of the present invention are described in further detail below.

In order to solve the problems in the background technology, the invention aims to firstly preliminarily determine a stereoscopic vision saliency map of a distorted stereoscopic image by simulating the visual saliency, and then optimize the preliminarily determined stereoscopic vision saliency map by simulating the characteristics of central offset and fovea, so as to extract a more accurate stereoscopic vision saliency map; secondly, fusing the quality maps of the left and right views and the quality map of the absolute difference map to obtain a comprehensive quality map of the stereo image; and finally, performing weighted summation on the stereo image comprehensive quality image by using the stereo saliency map to obtain an objective evaluation value of the quality of the distorted stereo image, which is described in detail below.

Example 1

A stereoscopic image quality objective evaluation method based on stereoscopic vision significance is disclosed, and referring to FIG. 11, the stereoscopic image quality objective evaluation method comprises the following steps:

101: optimizing the stereoscopic vision saliency map by simulating central deviation and fovea characteristics, and extracting the optimized stereoscopic vision saliency map;

102: obtaining a comprehensive quality image of the stereo image according to the quality images of the distorted stereo image and the absolute difference image;

103: and carrying out weighted summation on the stereo image comprehensive quality image through the stereo saliency map to obtain an objective evaluation value of the quality of the distorted stereo image.

Before step 101, the method for objectively evaluating the quality of a stereoscopic image further includes: the stereoscopic vision saliency map of the distorted stereoscopic image is obtained by simulating the visual saliency, and the steps are as follows:

converting the right view of the distorted stereo image from an RGB color space to a YCbCr color space, dividing the distorted stereo image into non-overlapping image blocks, and performing Discrete Cosine Transform (Discrete Cosine Transform) on the image blocks;

and normalizing the brightness contrast characteristic saliency map, the chroma contrast characteristic saliency map, the texture contrast characteristic saliency map and the absolute difference map of the distorted stereo image to obtain the stereo visual saliency map of the distorted stereo image.

In step 101, the stereoscopic visual saliency map is optimized by simulating the central shift and fovea characteristics, and the step of extracting the optimized stereoscopic visual saliency map specifically includes:

simulating a central offset factor, and performing primary optimization on the stereoscopic vision saliency map;

and simulating the foveal characteristic, and finally optimizing the preliminarily optimized stereoscopic vision saliency map.

The step 102 of obtaining the stereo image comprehensive quality map according to the quality maps of the distorted stereo image quality map and the absolute difference map specifically includes:

extracting the phase similarity of the right view of the distorted stereo image;

extracting gradient similarity of the original stereo image right view and the distorted stereo image right view;

respectively extracting the similarity of the chromaticity of the distorted right view;

calculating the similarity between the original stereo image right view and the distorted stereo image right view;

calculating the similarity of a left view of the distorted stereo image, and obtaining a quality image of the distorted stereo image by using the gradient similarity of a right view of the distorted stereo image and the similarity of the left view by using an average value method;

extracting a quality diagram of the absolute difference diagram of the distorted stereo image by adopting a structural similarity algorithm for the absolute difference diagram of the original stereo image and the absolute difference diagram of the distorted stereo image;

and fusing the quality image of the distorted stereo image with the quality image of the absolute difference image to obtain a comprehensive quality image of the stereo image.

In summary, the method realizes objective evaluation of the stereo image quality through the steps 101 to 103, accurately and effectively establishes an objective evaluation model of the stereo image quality reflecting subjective evaluation results, and meets various requirements in practical application.

Example 2

201: acquiring a stereoscopic vision saliency map of the distorted stereoscopic image by simulating the visual saliency;

according to the document [9], the method extracts the brightness contrast characteristic, the chromaticity contrast characteristic and the texture contrast characteristic of the right view of the distorted stereo image.

(1) Converting the right view of the distorted stereo image from RGB to YCbCr color space, dividing the distorted stereo image into non-overlapping image blocks, and performing discrete cosine transform on the image blocks;

converting the right view of the distorted stereo image from an RGB color space to a YCbCr color space, wherein a Y channel represents brightness information, and Cb and Cr channels represent chrominance information; dividing the right view of the distorted stereo image into 8-by-8 non-overlapping image blocks, and performing Discrete Cosine Transform (DCT) on the image blocks; direct Current (DC) coefficient L ═ Y with Y channel_DCDirect Current (DC) coefficient Cb of Cb channel representing brightness characteristics_DCDirect Current (DC) coefficient of Cr channel_DCRespectively representing the chrominance characteristics C₁And chroma feature C₂I.e. C₁＝Cb_DCAnd C₂＝Cr_DC(ii) a First 9 low frequency Alternating Current (AC) coefficients { Y) with Y channel_AC1,Y_AC2,...,Y_AC9Denotes a texture feature T, i.e., T ═ Y_AC1,Y_AC2,...,Y_AC9}。

Y＝0.299R+0.578G+0.114B

Cr＝0.5R-0.4178G-0.0813B+128 (1)

Cb＝-0.1687R-0.3313G+0.5B+128

The detailed operation steps of performing discrete cosine transform on an image block are well known to those skilled in the art, for example, refer to the transform method in document [10], which is not described in detail in the embodiments of the present invention.

(2) Calculating the significant value of the image block i to obtain a brightness contrast characteristic significant image F^LColor contrast feature saliency mapAndtexture contrast feature saliency map T;

the calculation process is well known to those skilled in the art, and for example, each feature saliency map may be obtained by a calculation method in reference [9], which is not described in detail in the embodiment of the present invention.

(3) Feature saliency map F^L、Andnormalizing the absolute difference map delta of the T and the distorted stereo image, and preliminarily obtaining a stereo saliency map SM1 by using a formula (2):

<math> <mrow> <mi>SM</mi> <mn>1</mn> <mo>=</mo> <mrow> <mo>(</mo> <msup> <mi>F</mi> <mi>L</mi> </msup> <msup> <mrow> <mo>+</mo> <mi>F</mi> </mrow> <msub> <mi>C</mi> <mn>1</mn> </msub> </msup> <mo>+</mo> <msup> <mi>F</mi> <msub> <mi>C</mi> <mn>2</mn> </msub> </msup> <mo>+</mo> <mi>T</mi> <mo>+</mo> <mi>&Delta;</mi> <mo>)</mo> </mrow> <mo>/</mo> <mn>5</mn> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </mrow> </math>

where Δ ═ L-R |, L and R are the left and right views of the distorted stereoscopic image, respectively.

202: optimizing the stereoscopic vision saliency map by simulating central deviation and fovea characteristics, and extracting the optimized stereoscopic vision saliency map;

(1) simulating a central offset factor, and performing primary optimization on the stereoscopic vision saliency map;

<math> <mrow> <mi>C</mi> <mi>B</mi> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>exp</mi> <mo>{</mo> <mo>-</mo> <mrow> <mo>(</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>i</mi> <mo>-</mo> <msub> <mi>i</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&sigma;</mi> <mi>h</mi> <mn>2</mn> </msubsup> </mrow> </mfrac> <mo>+</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>j</mi> <mo>-</mo> <msub> <mi>j</mi> <mn>0</mn> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&sigma;</mi> <mi>v</mi> <mn>2</mn> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>}</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>3</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein CB (i, j) represents the pixel point (i, j) of the distorted stereo image to the center point (i)₀,j₀) Is offset fromInformation; (i)₀,j₀) The coordinates of the central point of the distorted right view are expressed, and (i, j) is the x coordinate of the pixel point, sigma_hAnd σ_vThe standard deviations in the horizontal and vertical directions of the distorted stereoscopic image are expressed respectively and defined as σ_h＝W/3,σ_vW and H denote the number of horizontal pixels and vertical pixels of the distorted stereoscopic image, H/3.

The stereo-visual saliency map SM1 was initially optimized using equation (4):

SM'＝αSM1+(1-α)CB (4)

CB is a center shift map obtained by the formula (3), and α is a control parameter and defined as α being 0.7.

(2) And simulating the foveal characteristic, and finally optimizing the preliminarily optimized stereoscopic vision saliency map.

The foveal characteristic is modeled using equation (5):

<math> <mrow> <mi>C</mi> <mrow> <mo>(</mo> <mi>f</mi> <mo>,</mo> <mi>e</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mn>1</mn> <mrow> <msub> <mi>C</mi> <mn>0</mn> </msub> <mi>exp</mi> <mrow> <mo>(</mo> <mi>&delta;</mi> <mi>f</mi> <mo>(</mo> <mrow> <msub> <mi>e</mi> <mn>2</mn> </msub> <mo>+</mo> <mi>e</mi> </mrow> <mo>)</mo> <mo>/</mo> <msub> <mi>e</mi> <mn>2</mn> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>5</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein, the unit of e is degree (degree); f is the spatial frequency, in units ofCycle/degree (cycle/degree); c₀Is a contrast threshold; is a spatial frequency attenuation parameter; e.g. of the type₂Is the half resolution centrifugation constant. Fitting results from experiments [10]]Taken as 0.106, e₂＝2.3，C₀＝1/64。

And determining a threshold value by using a maximum inter-class variance method, and dividing the primarily optimized stereoscopic vision saliency map into a saliency region and a non-saliency region. Assume any pixel coordinate is (i)₁,j₁) The coordinate of the salient pixel point nearest to the salient pixel point is (i)₂,j₂) Then, retinal centrifugation degree e:

$e={\tan }^{-1}\left(\frac{d}{Wv}\right)---\left(6\right)$

wherein v is the viewing distance, the value of the method is 5 (unit is the image width), d is the pixel point (i)₁,j₁) And pixel point (i)₂,j₂) Euclidean distance of (c): $d=\sqrt{{(i_1-i_2)}^2+{(j_1-j_2)}^2}.$

and (3) optimizing the preliminarily optimized stereoscopic vision saliency map SM 'by using a formula (7) to obtain a final stereoscopic vision saliency map SM', and expanding the SM to be the same as the size of the right view of the distorted stereoscopic image.

SM＝SM'×C(f,e) (7)

203: obtaining a comprehensive quality image of the stereo image according to the quality images of the distorted stereo image and the absolute difference image;

the method combines the phase consistency characteristic, the image gradient characteristic and the chromaticity characteristic, respectively and independently processes the brightness information and the chromaticity information of the image in the YIQ color space, and adopts a characteristic similarity algorithm [11] to obtain the quality diagram of the distorted stereo image single view. Where Y represents luminance information of the image and I and Q represent chrominance information of the image, taking the right view as an example.

(1) Extracting phase similarity S of distorted stereo image right view_PC(x)；

Namely, using Reter Kovesi [12]]Method for extracting phase consistency characteristic PC of distorted stereo image right view at pixel point x by algorithm_d(x) As in equation (8). The same method is adopted to obtain the phase consistency characteristic PC of the original stereo image right view at the pixel point x_o(x)。

<math> <mrow> <msub> <mi>PC</mi> <mi>d</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&Sigma;</mo> <mi>l</mi> </munder> <msub> <mi>E</mi> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mo>&Sigma;</mo> <mi>l</mi> </munder> <munder> <mo>&Sigma;</mo> <mi>n</mi> </munder> <msub> <mi>A</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&theta;</mi> <mi>l</mi> </msub> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <mi>&epsiv;</mi> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>8</mn> <mo>)</mo> </mrow> </mrow> </math>

Wherein, theta_lL pi/L, L0, 1, L-1 is the filter's orientation angle,

l is the number of directions and is defined as L-4,is a direction theta_lAnd the amplitude of the response vector at the scale n,is a direction theta_lThe local energy function of (a) is a very small constant.

Determining the phase similarity of the distorted stereoscopic image right view by using the formula (9):

<math> <mrow> <msub> <mi>S</mi> <mrow> <mi>P</mi> <mi>C</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mn>2</mn> <msub> <mi>PC</mi> <mi>o</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <msub> <mi>PC</mi> <mi>d</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>T</mi> <mn>1</mn> </msub> </mrow> <mrow> <msub> <mi>PC</mi> <mi>o</mi> </msub> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>+</mo> <msub> <mi>PC</mi> <mi>d</mi> </msub> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>+</mo> <msub> <mi>T</mi> <mn>1</mn> </msub> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>9</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein, PC_o(x) And PC_d(x) Respectively at pixel points for original stereo image right view and distorted stereo image right view_xThe phase consistency value of (d). T is₁To prevent constants where the denominator is zero.

(2) Extracting gradient similarity of a right view of the distorted stereo image;

adopting a Scharr operator to respectively filter the brightness information Y of the original stereo image right view and the distorted stereo image right view, wherein the filtering values of the pixel points x in the horizontal direction and the vertical direction of the distorted stereo image right view are G respectively₁(x) And G₂(x) Calculating the pixel point of the distorted stereo image right view by using the formula (10)_xGradient amplitude GM of_d(x) Obtaining the gradient amplitude GM of the original stereo image right view at the pixel point x by the same method_o(x)。

${\mathrm{GM}}_d\left(x\right)=\sqrt{{\left(G_1\right(x\left)\right)}^2+{\left(G_2\right(x\left)\right)}^2}---\left(10\right)$

Calculating the gradient similarity of the distorted stereoscopic image right view by using a formula:

<math> <mrow> <msub> <mi>S</mi> <mrow> <mi>G</mi> <mi>M</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mn>2</mn> <msub> <mi>GM</mi> <mi>o</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <msub> <mi>GM</mi> <mi>d</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>T</mi> <mn>2</mn> </msub> </mrow> <mrow> <msub> <mi>GM</mi> <mi>o</mi> </msub> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>+</mo> <msub> <mi>GM</mi> <mi>d</mi> </msub> <msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>+</mo> <msub> <mi>T</mi> <mn>2</mn> </msub> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>11</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein, GM_o(x) And GM_d(x) The gradient amplitude values T of the original stereo image right view and the distorted stereo image right view at the pixel point x are respectively₂Is constant, and T₁Have the same meaning.

(3) Respectively extracting the similarity of the chroma I and the chroma Q of the distorted right view;

$S_I\left(x\right)=\frac{2I_o\left(x\right)I_d\left(x\right)+T_3}{{I_o}^2\left(x\right)+{I_d}^2\left(x\right)+T_3}---\left(12\right)$

$S_Q\left(x\right)=\frac{2Q_o\left(x\right)Q_d\left(x\right)+T_4}{{Q_o}^2\left(x\right)+{Q_d}^2\left(x\right)+T_4}---\left(13\right)$

wherein, T₃And T₄Is constant, and T₁Have the same meaning. I is_o、Q_oFor chrominance information of the right view of the original stereoscopic image, I_d、Q_dRespectively chrominance information of a right view of the distorted stereoscopic image.

(4) Calculating the similarity between the original stereo image right view and the distorted stereo image right view;

s is expressed by the following formula_pc(x)、S_GM(x)、S_I(x) And S_Q(x) And combining, calculating the similarity between the original stereo image right view and the distorted stereo image right view:

S_R(x)＝[S_pc(x)]^τ[S_GM(x)]^β[S_I(x)S_Q(x)]^λ (14)

wherein tau, beta and lambda are regulating parameters, and tau, beta and lambda are all defined as 1.

(5) Calculating the similarity S of the left view of the distorted stereo image according to the method for calculating the feature similarity of the right view of the distorted stereo image established in the steps (1) to (4)_L(x) Will S_R(x) And S_L(x) Obtaining a quality map S of a distorted stereo image by using an averaging method₀(x)。

(6) The structural similarity [1] is adopted for the absolute difference map of the original stereo image and the absolute difference map of the distorted stereo image](SSIM) algorithm, extracting the quality map d (x) of the absolute difference map of the distorted stereo image. Then, AlexandreBenoit [4] was used]The proposed algorithm is to distort the stereo image quality map S₀(x) And (3) fusing with the quality map D (x) of the absolute difference map to obtain a comprehensive quality map of the stereo image:

Q₀(x)＝S₀(x)·(1+D(x)) (15)

204: and carrying out weighted summation on the stereo image comprehensive quality image through the stereo saliency map to obtain an objective evaluation value of the quality of the distorted stereo image.

Stereoscopic saliency map SM of distorted stereoscopic image and stereoscopic imageComprehensive quality chart Q₀(x) And weighting, summing and normalizing to obtain the objective evaluation value Q of the quality of the stereo image.

<math> <mrow> <mi>Q</mi> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&Sigma;</mo> <mrow> <mi>x</mi> <mo>&Element;</mo> <mi>&Omega;</mi> </mrow> </munder> <msub> <mi>Q</mi> <mn>0</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&CenterDot;</mo> <mi>S</mi> <mi>M</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mo>&Sigma;</mo> <mrow> <mi>x</mi> <mo>&Element;</mo> <mi>&Omega;</mi> </mrow> </munder> <mi>S</mi> <mi>M</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>16</mn> <mo>)</mo> </mrow> </mrow> </math>

Where Ω represents the entire spatial domain of the image, and SM (x) is the value of the stereoscopic saliency map SM at pixel point x.

In summary, the method realizes objective evaluation of the stereo image quality through the steps 201 to 204, accurately and effectively establishes an objective evaluation model of the stereo image quality reflecting subjective evaluation results, and meets various requirements in practical application.

Example 3

The method is explained in detail below with reference to the technical scheme: the method specifically comprises three parts of extracting a stereoscopic vision saliency map, extracting a stereoscopic image comprehensive quality map and fusing the stereoscopic vision saliency map and the stereoscopic image comprehensive quality map, and specifically comprises the following steps:

301: preliminarily extracting a stereoscopic vision saliency map;

(1) converting the distorted stereoscopic image right view from an RGB color space to a YCbCr (4:2:0) color space according to formula (1), the Y channel representing luminance information, and the Cb and Cr channels representing chrominance information; dividing the right view into 8-by-8 non-overlapping image blocks, and performing discrete cosine transform on the image blocks according to a formula (17);

Y＝0.299R+0.578G+0.114B

Cr＝0.5R-0.4178G-0.0813B+128

Cb＝-0.1687R-0.3313G+0.5B+128

<math> <mrow> <mi>F</mi> <mrow> <mo>(</mo> <mrow> <mi>m</mi> <mo>,</mo> <mi>n</mi> </mrow> <mo>)</mo> </mrow> <mo>=</mo> <mi>c</mi> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mi>c</mi> <mrow> <mo>(</mo> <mi>n</mi> <mo>)</mo> </mrow> <munderover> <mo>&Sigma;</mo> <mrow> <mi>a</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>N</mi> <mo>-</mo> <mn>1</mn> </mrow> </munderover> <mrow> <munderover> <mo>&Sigma;</mo> <mrow> <mi>b</mi> <mo>=</mo> <mn>0</mn> </mrow> <mrow> <mi>N</mi> <mo>-</mo> <mn>1</mn> </mrow> </munderover> <mrow> <mi>f</mi> <mrow> <mo>(</mo> <mrow> <mi>a</mi> <mo>,</mo> <mi>b</mi> </mrow> <mo>)</mo> </mrow> <mi>cos</mi> <mfrac> <mrow> <mrow> <mo>(</mo> <mrow> <mn>2</mn> <mi>a</mi> <mo>+</mo> <mn>1</mn> </mrow> <mo>)</mo> </mrow> <mi>m</mi> <mi>&pi;</mi> </mrow> <mrow> <mn>2</mn> <mi>N</mi> </mrow> </mfrac> <mi>cos</mi> <mfrac> <mrow> <mrow> <mo>(</mo> <mrow> <mn>2</mn> <mi>b</mi> <mo>+</mo> <mn>1</mn> </mrow> <mo>)</mo> </mrow> <mi>n</mi> <mi>&pi;</mi> </mrow> <mrow> <mn>2</mn> <mi>N</mi> </mrow> </mfrac> </mrow> </mrow> <mo>,</mo> <mi>F</mi> <mo>&Element;</mo> <mrow> <mo>{</mo> <mrow> <mi>Y</mi> <mo>,</mo> <mi>C</mi> <mi>b</mi> <mo>,</mo> <mi>C</mi> <mi>r</mi> <mo>,</mo> <mi>T</mi> </mrow> <mo>}</mo> </mrow> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>17</mn> <mo>)</mo> </mrow> </mrow> </math>

<math> <mrow> <mtable> <mtr> <mtd> <mrow> <mi>c</mi> <mrow> <mo>(</mo> <mi>m</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = '{' close = ''> <mtable> <mtr> <mtd> <mrow> <msqrt> <mrow> <mn>1</mn> <mo>/</mo> <mi>N</mi> </mrow> </msqrt> <mo>,</mo> <mi>m</mi> <mo>=</mo> <mn>0</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msqrt> <mrow> <mn>2</mn> <mo>/</mo> <mi>N</mi> </mrow> </msqrt> <mo>,</mo> <mi>m</mi> <mo>&NotEqual;</mo> <mn>0</mn> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow> </mtd> <mtd> <mrow> <mi>c</mi> <mrow> <mo>(</mo> <mi>n</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = '{' close = ''> <mtable> <mtr> <mtd> <mrow> <msqrt> <mrow> <mn>1</mn> <mo>/</mo> <mi>N</mi> </mrow> </msqrt> <mo>,</mo> <mi>n</mi> <mo>=</mo> <mn>0</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <msqrt> <mrow> <mn>2</mn> <mo>/</mo> <mi>N</mi> </mrow> </msqrt> <mo>,</mo> <mi>n</mi> <mo>&NotEqual;</mo> <mn>0</mn> </mrow> </mtd> </mtr> </mtable> </mfenced> </mrow> </mtd> </mtr> </mtable> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>18</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein m and N are both [0, N-1 ]]And N is 8. F (m, n) represents a coefficient after discrete cosine transform. In the YCbCr color space, Y (m, n), Cb (m, n), and Cr (m, n) respectively represent coefficients of luminance information Y, chrominance information Cb, and Cr after discrete cosine transform. Calculating a Direct Current (DC) coefficient when m-n-0Further, the brightness characteristic L ═ Y is obtained_DCAnd colorimetric characteristics. The remaining, i.e. 63 coefficients, are Alternating Current (AC) coefficients. Acquiring first 9 low-frequency Alternating Current (AC) coefficients { Y (0,1), Y (1,0), Y (2,0), Y (1,1), Y (0,2), Y (0,3), Y (1,2), Y (2,1), Y (3,0) } of the channels by using a zig-zag scanning method, and further obtaining texture characteristics T ═ Y { Y (1,1) }_AC1,Y_AC2,...,Y_AC9}。

(2) Calculating the contrast of the image block i and the rest image blocks by using a formula (19) to estimate the saliency of the image block i to obtain a brightness contrast characteristic saliency map F^LColor contrast feature saliency mapAndtexture contrast feature saliency map T.

<math> <mrow> <msubsup> <mi>F</mi> <mi>i</mi> <mi>k</mi> </msubsup> <mo>=</mo> <munder> <mo>&Sigma;</mo> <mrow> <mi>i</mi> <mo>&NotEqual;</mo> <mi>j</mi> </mrow> </munder> <mfrac> <mn>1</mn> <mrow> <mi>&sigma;</mi> <msqrt> <mrow> <mn>2</mn> <mi>&pi;</mi> </mrow> </msqrt> </mrow> </mfrac> <msup> <mi>e</mi> <mrow> <mo>-</mo> <msubsup> <mi>l</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mn>2</mn> </msubsup> <mo>/</mo> <mrow> <mo>(</mo> <mn>2</mn> <msup> <mi>&sigma;</mi> <mn>2</mn> </msup> <mo>)</mo> </mrow> </mrow> </msup> <msubsup> <mi>U</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mi>k</mi> </msubsup> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>19</mn> <mo>)</mo> </mrow> </mrow> </math>

Wherein k represents a feature and k ∈ { L, C₁,C₂T, σ is a parameter and is defined as 20, l_ijBeing the euclidean distance between image blocks i and j,representing the feature contrast of image blocks i and j over feature k.

The contrast of the brightness characteristic and the chroma characteristic is calculated by the formula (20), and the contrast of the texture characteristic is calculated by the formula (21)

<math> <mrow> <msubsup> <mi>U</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mi>d</mi> </msubsup> <mo>=</mo> <mfrac> <mrow> <mo>|</mo> <mrow> <msubsup> <mi>B</mi> <mi>i</mi> <mi>d</mi> </msubsup> <mo>-</mo> <msubsup> <mi>B</mi> <mi>j</mi> <mi>d</mi> </msubsup> </mrow> <mo>|</mo> </mrow> <mrow> <msubsup> <mi>B</mi> <mi>i</mi> <mi>d</mi> </msubsup> <mo>+</mo> <msubsup> <mi>B</mi> <mi>j</mi> <mi>d</mi> </msubsup> </mrow> </mfrac> <mo>,</mo> <mi>d</mi> <mo>&Element;</mo> <mo>{</mo> <mi>L</mi> <mo>,</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> <mo>,</mo> <msub> <mi>C</mi> <mn>2</mn> </msub> <mo>}</mo> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>20</mn> <mo>)</mo> </mrow> </mrow> </math>

<math> <mrow> <msubsup> <mi>U</mi> <mrow> <mi>i</mi> <mi>j</mi> </mrow> <mi>T</mi> </msubsup> <mo>=</mo> <mfrac> <msqrt> <mrow> <msub> <mi>&Sigma;</mi> <mi>t</mi> </msub> <msup> <mrow> <mo>(</mo> <msubsup> <mi>B</mi> <mi>i</mi> <mrow> <mi>T</mi> <mi>t</mi> </mrow> </msubsup> <mo>-</mo> <msubsup> <mi>B</mi> <mi>j</mi> <mrow> <mi>T</mi> <mi>t</mi> </mrow> </msubsup> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </msqrt> <mrow> <msub> <mi>&Sigma;</mi> <mi>t</mi> </msub> <mrow> <mo>(</mo> <msubsup> <mi>B</mi> <mi>i</mi> <mrow> <mi>T</mi> <mi>t</mi> </mrow> </msubsup> <mo>+</mo> <msubsup> <mi>B</mi> <mi>j</mi> <mrow> <mi>T</mi> <mi>t</mi> </mrow> </msubsup> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>21</mn> <mo>)</mo> </mrow> </mrow> </math>

Wherein,andthe characteristic values of the image blocks i and j on the characteristic d, t is an Alternating Current (AC) coefficient

t∈{1,2,2,...,9}，Representing the coefficients T of the image block i on the texture feature T.

(3) Using the feature saliency map F obtained by the formula (19)^L,And (3) carrying out Max-Min normalization operation on the T and the absolute difference diagram delta of the distorted stereo image, and preliminarily obtaining a stereo saliency map SM1 according to a formula (22):

<math> <mrow> <mi>S</mi> <mi>M</mi> <mn>1</mn> <mo>=</mo> <mrow> <mo>(</mo> <msup> <mi>F</mi> <mi>L</mi> </msup> <mo>+</mo> <msup> <mi>F</mi> <msub> <mi>C</mi> <mn>1</mn> </msub> </msup> <mo>+</mo> <msup> <mi>F</mi> <msub> <mi>C</mi> <mn>2</mn> </msub> </msup> <mo>+</mo> <mi>T</mi> <mo>+</mo> <mi>&Delta;</mi> <mo>)</mo> </mrow> <mo>/</mo> <mn>5</mn> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>22</mn> <mo>)</mo> </mrow> </mrow> </math>

wherein, L and R are gray level images of a left view and a right view of the distorted stereo image respectively

302: optimizing the stereoscopic saliency map;

(1) the central offset factor [13] was simulated and the saliency map SM1 was optimized.

(2) Retinal photoreceptor density decreases rapidly from the fovea to the periphery [14][15]. Therefore, when the image is mapped on the retina, the spatial frequency resolution of the human eye to the foveal region is higher, and the partial region is the region focused by the human eye, namely the salient region; as the eccentricity e increases, the spatial resolution of the human eye decreases. Using Contrast Sensitivity Function (CSF) [15 ]]This property was simulated, based on experimental fitting results [10]]Taken as 0.106, e₂＝2.3,C₀＝1/64。

And determining a threshold value by using a maximum inter-class variance method, and dividing the stereoscopic saliency map into a salient region and a non-salient region. And multiplying the points SM 'and C (f, e) to optimize the stereoscopic vision saliency map SM' to obtain a final stereoscopic vision saliency map, and expanding the stereoscopic vision saliency map to be the same as the size of the right view of the distorted stereoscopic image.

The specific operation of this step is described in embodiment 2, which is not described in detail in this embodiment of the present invention.

303: a quality map of the distorted stereoscopic image is calculated.

The method combines the phase consistency characteristic, the image gradient characteristic and the chromaticity characteristic, respectively and independently processes the brightness information and the chromaticity information of the image in the YIQ color space, and adopts a characteristic similarity algorithm [11] to obtain the quality diagram of a distorted stereo image single view, taking a right view as an example.

(1) The image is converted from RGB color space to YIQ color space, and the phase consistency information of original and distorted right view at pixel point x is extracted by using the Reter Kovesi [12] algorithm. The phase similarity of the distorted right view is determined.

(2) Gradient similarities of the original and distorted right views are extracted.

(3) And extracting the chroma similarity of the distorted right view.

(4) And calculating the similarity between the original stereo image right view and the distorted right view.

(5) Calculating the similarity S of the distorted left view according to the method for calculating the feature similarity of the distorted right view established in the steps (1) to (4)_L(x)。

(6) The structural similarity [1] is adopted for the absolute difference map of the original stereo image and the absolute difference map of the distorted stereo image](SSIM) algorithm, extracting the quality map d (x) of the absolute difference map of the distorted stereo image. With Alexandre Benoit [4]]Algorithm, image quality map S₀(x) Fusing with the quality map D (x) of the absolute difference map to obtain a comprehensive quality map Q of the stereo image₀(x)。

The specific operation of this step is described in embodiment 2, which is not described in detail in this embodiment of the present invention.

304: and obtaining the objective evaluation value of the quality of the stereo image.

And weighting and summing the stereoscopic vision saliency map SM and the stereoscopic image quality map of the distorted stereoscopic image and normalizing to obtain the objective evaluation value of the stereoscopic image quality.

In summary, the method realizes objective evaluation of the stereo image quality through steps 301 to 304, accurately and effectively establishes an objective evaluation model of the stereo image quality reflecting subjective evaluation results, and meets various requirements in practical application.

Example 4

The reference experimental materials used in the design are all taken from a stereo information base provided by a broadband wireless communication and stereo imaging research institute, and 8 original images which are not compressed, noised and blurred in the image base are selected, and are respectively 'tree 2', 'family', 'girl', 'river', 'tree 1', 'ox', 'tju' and 'wman', and are shown in fig. 1-8. And (3) carrying out JPEG compression distortion, Gaussian blur distortion and Gaussian noise distortion processing on the 8 standard stereo image pairs to obtain 260 distorted stereo image pairs. Each image has a resolution of 2560 x 1024. Subjective evaluation was performed on all distorted images and mean opinion values (MOS) were recorded. In 406 stereo image samples, 150 images are selected as training samples, and 256 images are selected as test samples.

Fig. 10 is a schematic diagram comparing before and after optimization of the stereoscopic saliency map, fig. 10(a) is a right view of a distorted stereoscopic image, (b) is a stereoscopic saliency map SM1 before optimization, (c) is a stereoscopic saliency map SM' after being optimized for center shift, and (d) is a stereoscopic saliency map SM after being optimized for fovea. Comparing (b) to (c) finds that the pixels in the central region of the image have increased significance after the central shift factor is introduced, as pixels closer to the central region of the image are more likely to draw the attention of the viewer due to the presence of the central shift factor; under the influence of the foveal characteristic, the human eye has low resolution for regions further from the salient region, so the saliency for the pixels of the non-salient region after (c) optimization (d) is reduced. Therefore, the stereoscopic saliency map optimized by the central offset factor and the foveal characteristic better conforms to the visual characteristics of human eyes.

Table 1 lists the performance indexes between the subjective and objective evaluation values of the algorithm and SSIM [1], DSSIM [3] and 3SIQA [8 ].

TABLE 1 Performance indices for different evaluation algorithms

As can be seen from the data in Table 1, for different distortion types, the PLCC index and the RMSE index of the algorithm are improved to different degrees, the PLCC value is above 0.95, and the RMSE value is below 0.45. Although the KROCC value of the algorithm for Gaussian noise distortion and Gaussian blur distortion is slightly lower than that of the algorithm in the document [8], the overall performance index of the algorithm is far higher than that of the other three algorithms, because the depth information is not considered in the algorithms in the documents [8] and [1], the influence of the stereoscopic vision significance is not considered in the algorithms in the documents [3] and [1], and the influence of the stereoscopic vision significance and the depth information is considered in the algorithm, the objective evaluation value and the subjective evaluation result of the algorithm have better consistency aiming at different distortion types and distortion intensities.

Reference to the literature

[1]Z.Wang,A.C.Bovik A,H.R.Sheikh,et al.Image Quality Assessment:From Error Visibilityto Structural Similarity[J].IEEE Transactions on Image Processing,Vol.13,No.4,April 2004.

[2]Patrizio Campisi,Patrick Le Callet,Enrico Marini.STEREOSCOPIC IMAGES QUALITYASSESSMENT[C].15th European Signal Processing Conference(EUSIPCO 2007),Poznan,Poland,September 3-7,2007.

[3]de Miranda Cardoso,J.V.,Danilo Miranda Regis,C.,Sampaio de Alencar,M.On thepreference of disparity-based weighting technique applied to 3D image quality assessment[C].Broadband Multimedia Systems and Broadcasting(BMSB),2014 IEEE internationalSymposium on,Beijing,25-27June 2014.

[4]Alexandre Benoit,Patrick Le Callet,Patrizio Campisi,et al.Using disparity for qualityassessment of stereoscopic images[C].15th IEEE International Conference on Image Processing(ICIP).San Diego,CA,2008:389-392.

[5]Dongdong Zhang,Jiahe Huang,Di Zang,et al.An quality metric for 3D rendered image basedon stereo saliency and structural similarity[C].Global High Tech Congress on Electronics(GHTCE),2013IEEE,Shenzhen,17-19Nov.2013:170-174.

[6]Qi F,Jiang T T,Ma S W,et al.Quality of experience assessment for stereoscopic images[C].2012IEEEInternationalSymposiumonCirruitsandSystems(ISCAS),Seoul,Korea,2012:1712–1715.[7]Shao Feng,Duan Fen-fang,et al.3D Visual attention for stereoscopic imagequalityassessment[J].Journal of software,2014,Vol 9,No 7:1841-1847.

[8]Xiu-qin Chu,Yuan-Yuan Wu,Qian Li.Saliency structure stereoscopic image quality assessmentmethod[J].Optik-International Journal for Light and Electron Optics,2014,125(2):704-709.

[9]Yuming Fang,Junle Wang,Manish Narwaria,et al.Saliency Detection for StereoscopicImages[J].IEEE TRANSACTIONS ON IMAGE PROCESSING,VOL.23,NO.6,JUNE 2014.

[10]W.S.Geisler and J.S.Perry,A real-time foveated multiresolution system for low-bandwidthvideo communication,Proc.SPIE,vol.3299,pp.294–305,July 1998.

[11]Lin Zhang,Lei Zhang,Xuanqin Mou,et al.FSIM:A Feature Similarity Index for ImageQuality Assessmen[J].IEEE TRANSACTIONS ON IMAGE PROCESSING,VOL.20,NO.8,AUGUST 2011.

[12]Kovesi P.Image features from phase congruency[J].Videre:Journal of Computer VisionResearch,1999,1(3):1-30.

[13]Le Meur O.,Le Callet,P.,Barba,et al.A coherent computational approach to modelbottom-up visual attention[J].Pattern Analysis and Machine Intelligence,IEEE Transactions on,2006,28(5):802-817.

[14]P.Tseng,R.Carmi,I.G.M.Camerson,et al.Quantifying center bias of observers in freeviewing of dynamic natural scenes[J].Journal of Vision,2009,vol.9,no.7.

[15]Zhou Wang,Ligang Lu,Alan Conrad Bovik.Foveation Scalable Video Coding WithAutomatic Fixation Selection [J].IEEE TRANSACTIONS ON IMAGE PROCESSING,FEBRUARY 2003,VOL.12,NO.2:243-254.

In the embodiment of the present invention, except for the specific description of the model of each device, the model of other devices is not limited, as long as the device can perform the above functions.

Those skilled in the art will appreciate that the drawings are only schematic illustrations of preferred embodiments, and the above-described embodiments of the present invention are merely provided for description and do not represent the merits of the embodiments.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims (5)

1. A stereoscopic image quality objective evaluation method based on stereoscopic visual saliency is characterized by comprising the following steps:

optimizing the stereoscopic vision saliency map by simulating central deviation and fovea characteristics, and extracting the optimized stereoscopic vision saliency map;

obtaining a comprehensive quality image of the stereo image according to the quality images of the distorted stereo image and the absolute difference image;

and carrying out weighted summation on the stereo image comprehensive quality image through the stereo salient image to obtain an objective evaluation value of the quality of the distorted stereo image.

2. The objective evaluation method for stereoscopic image quality based on stereoscopic saliency according to claim 1, characterized in that before the step of extracting the optimized stereoscopic saliency map by optimizing the stereoscopic saliency map through simulating the center shift and fovea characteristics, the objective evaluation method for stereoscopic image quality further comprises the following steps:

and acquiring a stereoscopic vision saliency map of the distorted stereoscopic image by simulating the visual saliency.

3. The method for objectively evaluating the quality of a stereoscopic image based on stereoscopic saliency according to claim 1, wherein the step of obtaining the stereoscopic saliency map of the distorted stereoscopic image by simulating the visual saliency is specifically as follows:

converting the right view of the distorted stereo image from an RGB color space to a YCbCr color space, dividing the distorted stereo image into non-overlapping image blocks, and performing discrete cosine transform on the image blocks;

and normalizing the brightness contrast characteristic saliency map, the chroma contrast characteristic saliency map, the texture contrast characteristic saliency map and the absolute difference map of the distorted stereo image to obtain the stereo visual saliency map of the distorted stereo image.

4. The objective evaluation method for stereoscopic image quality based on stereoscopic saliency according to claim 1, characterized in that the stereoscopic saliency map is optimized by simulating the center shift and fovea characteristics, and the step of extracting the optimized stereoscopic saliency map comprises:

simulating a central offset factor, and performing primary optimization on the stereoscopic vision saliency map;

and simulating the foveal characteristic, and finally optimizing the preliminarily optimized stereoscopic vision saliency map.

5. The objective evaluation method for stereoscopic image quality based on stereoscopic visual saliency according to claim 1, wherein the step of obtaining the stereoscopic image comprehensive quality map from the quality maps of the distorted stereoscopic image quality map and the absolute difference map specifically comprises:

extracting the phase similarity of the right view of the distorted stereo image;

extracting gradient similarity of a right view of the distorted stereo image;

respectively extracting the similarity of the chromaticity of the distorted right view;

calculating the similarity between the original stereo image right view and the distorted stereo image right view;

calculating the similarity of a left view of the distorted stereo image, and obtaining a quality image of the distorted stereo image by using the gradient similarity of a right view of the distorted stereo image and the similarity of the left view by using an average value method;

extracting a quality diagram of the absolute difference diagram of the distorted stereo image by adopting a structural similarity algorithm for the absolute difference diagram of the original stereo image and the absolute difference diagram of the distorted stereo image;

and fusing the quality image of the distorted stereo image with the quality image of the absolute difference image to obtain a comprehensive quality image of the stereo image.