CN107705286A - A kind of color image quality integrated evaluating method - Google Patents

Abstract

The invention discloses a color image quality comprehensive evaluation method. Firstly, a color image to be evaluated and a corresponding reference image are input; the reference image and the image to be evaluated are respectively subjected to color space transformation, and brightness channel diagrams of the reference image and the image to be evaluated are obtained and the chromaticity channel map, extract the chromaticity channel map of the reference image and the image to be evaluated, and calculate the chromaticity similarity feature; extract the luminance channel map of the reference image and the image to be evaluated, and use log-Gabor wavelet to obtain the similarity feature of phase consistency; The positional saliency features of the reference image and the image to be evaluated are obtained; the quality of the image to be evaluated is obtained through the standard deviation pooled phase consistency similarity feature and chromaticity similarity feature weighted by the positional saliency feature. The invention fully considers the different characteristics of the human visual system, reasonably combines these three aspects, and better evaluates the quality of the color image.

Description

A Comprehensive Evaluation Method of Color Image Quality

technical field

The invention belongs to the field of image quality evaluation, in particular to a color image quality comprehensive evaluation method based on phase consistency similarity, chromaticity similarity and position saliency.

Background technique

At present, the full-reference image quality evaluation methods mainly used in the field of image quality evaluation are mean square error, peak signal-to-noise ratio method, etc. These methods only consider the pixel-by-pixel difference between the reference image and the image to be evaluated, and do not consider the human vision The characteristics of the system. Therefore these methods cannot reflect the subjective quality of the image well.

The human visual system is sensitive to edge structure information; and it is more sensitive to the distortion perception of the central area of the image than the edge area; the difference in different color spaces of color images also needs to be considered. However, some existing evaluation methods that consider human visual characteristics, such as structural similarity, lack comprehensive evaluation of these three types of characteristics.

Contents of the invention

Purpose of the invention: In order to solve the problems existing in the prior art and better reflect the subjective quality of images from the perspective of human visual characteristics, the present invention provides a comprehensive evaluation method for color image quality.

Technical solution: A method for comprehensive evaluation of color image quality provided by the present invention includes the following steps:

(1) Input the color image and reference image to be evaluated;

(2) Perform color space transformation on the reference image and the image to be evaluated respectively, obtain the luminance channel map and the chrominance channel map of the reference image and the image to be evaluated, extract the chroma channel map of the reference image and the image to be evaluated, and calculate the chromaticity similarity sexual characteristics;

(3) Extract the luminance channel maps of the reference image and the image to be evaluated, and use log-Gabor wavelet to obtain phase consistency similar features;

(4) According to the size of the reference image and the image to be evaluated, using the position saliency formula to obtain the positional saliency features of the reference image and the image to be evaluated;

(5) The quality of the image to be evaluated is obtained by standard deviation pooling phase consistency similarity feature and chromaticity similarity feature weighted by positional saliency feature.

Preferably, let x represent the pixel points in the reference image or the image to be evaluated, let the phase consistency similarity feature in step (3) be S _pc (x), let the chroma similarity feature in step (2) be S _c (x), if the positional saliency feature is S _d (x), the step (5) specifically includes:

(51) Let the overall similarity feature be SM(x), and combine S _pc (x) and S _c (x) to calculate SM(x):

SM(x)＝S _pc (x)·S _c (x)

(52) Calculate the weight of each point in the subsequent pooling quality feature by using the positional saliency feature size:

(53) Set the overall similarity quality value as SMD, and calculate the SMD to evaluate the quality of the image to be evaluated by pooling the phase consistency similarity and chromaticity similarity features with the weighted standard deviation of the positional salience feature:

Among them, MM and NN are the width and height of the image to be evaluated, respectively.

By combining phase consistency and chromaticity similarity, the method is suitable for the comprehensive evaluation of image structure and color information, and the scope of application of the evaluation method is improved; in addition, combined with the position saliency feature, the subjective consistency of the evaluation method is improved.

Preferably, the specific method for calculating the chromaticity similarity feature in the step (2) includes:

(21) The reference image and the image to be evaluated are respectively subjected to color space transformation:

Among them, R, G, and B respectively represent the red, green and blue channels of the color image; L represents the brightness channel, and M and N represent the chrominance channel;

(22) Extract the M and N chromaticity channel maps of the reference image and the image to be evaluated, and calculate the chromaticity similarity features of the reference image and the image to be evaluated pixel by pixel, and the calculation formula is:

S _c (x) = S _cm (x) · S _cn (x)

Wherein C ₁ is a normal number, M ₁ (x) and M ₂ (x) respectively represent the gray value of the M chroma channel of the reference image and the gray value of the M chroma channel of the image to be evaluated, N ₁ (x) and N ₂ ( x) represent the gray value of the M chroma channel of the reference image and the gray value of the N chroma channel of the image to be evaluated, S _cm (x) is the similarity feature of the M chroma channel, S _cn (x) is the similarity feature of the N chroma channel sexual characteristics.

Preferably, the specific method that described step (3) adopts log-Gabor wavelet to obtain phase consistency similar features includes:

(31) Using a two-dimensional log-Gabor filter, calculate the even symmetric filter response of the reference image and the image to be evaluated L with a brightness channel map point x in the direction _θj and scale n and odd symmetric filter response The two-dimensional log-Gabor filter expression is:

in j represents the jth direction, J represents the number of directions, σ _θ is used to determine the bandwidth of the filter angle; θ represents the direction angle of the filter; ω represents the angular frequency of the filter; ω ₀ represents the two-bit log-Gabor filter The center frequency of , k represents the shape parameter of the filter, and the shape of the filter is determined by ω ₀ and k.

(32) Calculating the sum magnitude under the direction θ _j and scale n with the response local energy in direction θ _j

(33) Synthesize the responses of all directions and scales, and calculate the phase consistency of each point:

Where ε is a normal number;

Using the same method to calculate the luminance channel phase consistency feature PC 1 (x) of the reference image and the luminance channel phase consistency feature PC ₂ (x) of the image to be evaluated _;

(34) Calculate the phase-consistency similarity features of the reference image and the image to be evaluated L luminance channel diagram pixel by pixel:

Among them, C ₂ is a normal constant.

Preferably, the positional significance characteristics of each point in the step (4) are:

Among them, x _c is the center point of the feature map; Indicates the distance between the feature map x point and the center point; σ _d ² is an empirical parameter.

Beneficial effects: the present invention provides a color image quality comprehensive evaluation method, which uses the similarity of image phase consistency to evaluate the perception characteristics of the human visual system for structural information; uses the position saliency of the image to evaluate the human visual system for spatial position Perceptual properties; evaluating color information of color images using their chromatic similarity. The method takes full account of the different characteristics of the human visual system, reasonably combines these three aspects, and better evaluates the quality of color images.

Description of drawings

Fig. 1 is a method flow chart of the comprehensive evaluation method for color image quality of the present invention.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

As shown in Figure 1, a color image quality comprehensive evaluation method includes the following steps:

(1) Input the color image to be evaluated and the reference image; let x represent the pixel in the reference image or the image to be evaluated. Among them, the reference image is the "perfect image" of the color image to be evaluated that is not affected by distortion (there is no unified quantification standard for this "undistorted" at present), and the reference image needs to be consistent with the size and number of channels of the image to be evaluated .

(2) Perform color space transformation on the reference image and the image to be evaluated respectively, obtain the luminance channel map and the chrominance channel map of the reference image and the image to be evaluated, extract the chroma channel map of the reference image and the image to be evaluated, and calculate the chromaticity similarity feature; let the chroma similarity feature be S _c (x), the method of obtaining S _c (x) is as follows:

(21) The reference image and the image to be evaluated are respectively subjected to color space transformation:

Among them, R, G, and B respectively represent the red, green and blue channels of the color image; L represents the brightness channel, and M and N represent the chrominance channel;

(22) Extract the M and N chromaticity channel maps of the reference image and the image to be evaluated, and calculate the chromaticity similarity features of the reference image and the image to be evaluated pixel by pixel, and the calculation formula is:

S _c (x) = S _cm (x) · S _cn (x)

Wherein C ₁ is a normal number, M ₁ (x) and M ₂ (x) respectively represent the gray value of the M chroma channel of the reference image and the gray value of the M chroma channel of the image to be evaluated, N ₁ (x) and N ₂ ( x) represent the gray value of the M chroma channel of the reference image and the gray value of the N chroma channel of the image to be evaluated, S _cm (x) is the similarity feature of the M chroma channel, S _cn (x) is the similarity feature of the N chroma channel sexual characteristics.

(3) Extract the luminance channel images of the reference image and the image to be evaluated, and use the log-Gabor wavelet to obtain the phase consistency similarity feature; set the phase consistency similarity feature as S _pc (x), and the method of obtaining S _pc (x) is as follows :

(31) Using a two-dimensional log-Gabor filter, calculate the even symmetric filter response of the reference image and the image to be evaluated L with a brightness channel map point x in the direction _θj and scale n and odd symmetric filter response The two-dimensional log-Gabor filter expression is:

in j represents the jth direction, J represents the number of directions, σ _θ is used to determine the bandwidth of the filter angle; θ represents the direction angle of the filter; ω represents the angular frequency of the filter; ω ₀ represents the two-bit log-Gabor filter The center frequency of , k represents the shape parameter of the filter, and the shape of the filter is determined by ω ₀ and k.

(32) Calculating the sum magnitude under the direction θ _j and scale n with the response local energy in direction θ _j

(33) Synthesize the responses of all directions and scales, and calculate the phase consistency of each point:

Where ε is a normal number;

Using the same method to calculate the luminance channel phase consistency feature PC 1 (x) of the reference image and the luminance channel phase consistency feature PC ₂ (x) of the image to be evaluated _;

(34) Calculate the phase-consistency similarity features of the reference image and the image to be evaluated L luminance channel diagram pixel by pixel:

Among them, C ₂ is a normal constant.

(4) According to the size of the reference image and the image to be evaluated, use the positional saliency formula to obtain the positional saliency features of the reference image and the image to be evaluated; set the positional saliency feature as S _d (x), and obtain S _d (x) The method is as follows:

Among them, x _c is the center point of the feature map; Indicates the distance between the feature map x point and the center point; σ _d ² is an empirical parameter.

(5) Obtain the quality of the image to be evaluated through the standard error pooling phase consistency similarity feature and chromaticity similarity feature weighted by the positional saliency feature, including:

(51) Let the overall similarity feature be SM(x), and combine S _pc (x) and S _c (x) to calculate SM(x):

SM(x)＝S _pc (x)·S _c (x)

(52) Calculate the weight of each point in the subsequent pooling quality feature by using the positional saliency feature size:

(53) Set the overall similarity quality value as SMD, and calculate the SMD to evaluate the quality of the image to be evaluated by pooling the phase consistency similarity and chromaticity similarity features with the weighted standard deviation of the positional salience feature:

Among them, MM and NN are the width and height of the image to be evaluated, respectively.

For step (51), since the image feature positions with high recognition degree in human vision, such as the edge point positions of the image, are consistent with those point positions with congruent phases of the frequency domain coefficients, the coefficients describing the frequency domain transformation of the signal The phase consistency feature of the phase angle consistency between them can effectively characterize the acceptance process of the human eye for the image structure information, but the phase consistency cannot extract the color information in the color image, so the combination of phase consistency similarity and chromaticity similarity , so that the extracted quality features can measure the structure and color information of the image more comprehensively.

For step (52), after obtaining the joint feature map of the phase consistency similarity and chromaticity similarity of the reference image and the distorted image, since the human visual system perceives the central area of the image area more than the surrounding edge area, the position The saliency feature map weights the quality features obtained in step (51), so that the quality features of the central region have a larger proportion than the surrounding regions, which is more in line with the perception of the human visual system.

For step (53), the overall evaluation value is obtained by pooling the weighted quality features. The pooling method of local quality features used in the usual image quality evaluation method is the average pooling method. When the average pooling method counts local features, it implies that each pixel of the image has the same importance to the overall image quality. Because the same degree of distortion causes different degrees of perceptual quality changes in different image regions, such as the same blurred distortion, it will cause greater perceptual quality changes in structural and textured regions than in smooth regions. The standard error pooling method can better reflect the range of image distortion than the average pooling method.

The method in step (51) makes the method suitable for the comprehensive evaluation of image structure and color information by combining phase consistency and chromaticity similarity, and improves the scope of application of the evaluation method; in addition, combined with the position saliency feature, the evaluation method is improved. Subjective consistency.

Claims (5)

1. A color image quality comprehensive evaluation method is characterized by comprising the following steps:

(1) inputting a color image to be evaluated and a reference image;

(2) respectively carrying out color space transformation on the reference image and the image to be evaluated to obtain a luminance channel image and a chrominance channel image of the reference image and the image to be evaluated, extracting the chrominance channel images of the reference image and the image to be evaluated, and calculating chrominance similarity characteristics;

(3) extracting brightness channel graphs of a reference image and an image to be evaluated, and obtaining phase consistency similar characteristics by using log-Gabor wavelets;

(4) according to the sizes of the reference image and the image to be evaluated, acquiring position significance characteristics of the reference image and the image to be evaluated by using a position significance formula;

(5) and obtaining the quality of the image to be evaluated through standard deviation pooling phase consistency similarity characteristics and chromaticity similarity characteristics weighted by the position saliency characteristics.

2. The color image quality comprehensive evaluation method according to claim 1, wherein x is set to represent a pixel point in a reference image or an image to be evaluated, and the phase consistency similarity characteristic in the step (3) is set to be S_pc(x) The chroma similarity characteristic in the step (2) is set as S_c(x) Let the location saliency be S_d(x) The step (5) specifically comprises:

(51) let overall similarity feature SM (x), in combination with S_pc(x) And S_c(x) Calculate sm (x):

SM(x)＝S_pc(x)·S_c(x)

(52) and calculating the weight of each point in the subsequent pooling quality characteristics by using the position significance characteristic size:

<mrow> <msub> <mi>S</mi> <mrow> <mi>r</mi> <mi>d</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <msub> <mi>S</mi> <mi>d</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mo>&amp;Sigma;</mo> <mi>x</mi> </munder> <msub> <mi>S</mi> <mi>d</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> </mfrac> </mrow>

(53) and (3) setting the total similar property quantity value as an SMD, calculating the SMD to evaluate the quality of the image to be evaluated by pooling phase consistency and chromaticity similarity characteristics through the position significance characteristic weighting standard difference:

<mrow> <mover> <mrow> <mi>S</mi> <mi>M</mi> </mrow> <mo>&amp;OverBar;</mo> </mover> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&amp;Sigma;</mo> <mi>x</mi> </munder> <mi>S</mi> <mi>M</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> <mrow> <mi>M</mi> <mi>M</mi> <mo>&amp;CenterDot;</mo> <mi>N</mi> <mi>N</mi> </mrow> </mfrac> </mrow>

<mrow> <mi>S</mi> <mi>M</mi> <mi>D</mi> <mo>=</mo> <msqrt> <mrow> <msub> <mi>S</mi> <mrow> <mi>r</mi> <mi>d</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&amp;CenterDot;</mo> <msup> <mrow> <mo>(</mo> <mi>S</mi> <mi>M</mi> <mo>(</mo> <mi>x</mi> <mo>)</mo> <mo>-</mo> <mover> <mrow> <mi>S</mi> <mi>M</mi> </mrow> <mo>&amp;OverBar;</mo> </mover> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </msqrt> </mrow>

wherein, MM and NN are respectively the width and height of the image to be evaluated.

3. The comprehensive color image quality evaluation method according to claim 2, wherein the specific method for calculating the chroma similarity characteristic in the step (2) comprises the following steps:

(21) and respectively carrying out color space transformation on the reference image and the image to be evaluated:

<mrow> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mi>L</mi> </mtd> </mtr> <mtr> <mtd> <mi>M</mi> </mtd> </mtr> <mtr> <mtd> <mi>N</mi> </mtd> </mtr> </mtable> </mfenced> <mo>=</mo> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mn>0.06</mn> </mtd> <mtd> <mn>0.63</mn> </mtd> <mtd> <mn>0.27</mn> </mtd> </mtr> <mtr> <mtd> <mn>0.30</mn> </mtd> <mtd> <mn>0.04</mn> </mtd> <mtd> <mrow> <mo>-</mo> <mn>0.35</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mn>0.34</mn> </mtd> <mtd> <mrow> <mo>-</mo> <mn>0.60</mn> </mrow> </mtd> <mtd> <mn>0.17</mn> </mtd> </mtr> </mtable> </mfenced> <mfenced open = "[" close = "]"> <mtable> <mtr> <mtd> <mi>R</mi> </mtd> </mtr> <mtr> <mtd> <mi>G</mi> </mtd> </mtr> <mtr> <mtd> <mi>B</mi> </mtd> </mtr> </mtable> </mfenced> </mrow>

wherein R, G, B represent the red, green, and blue channels of a color image, respectively; l denotes a luminance channel, M, N denotes a chrominance channel;

(22) extracting M, N chrominance channel images of the reference image and the image to be evaluated, and calculating the chrominance similarity characteristics of the reference image and the image to be evaluated pixel by pixel, wherein the calculation formula is as follows:

<mrow> <msub> <mi>S</mi> <mrow> <mi>c</mi> <mi>m</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mn>2</mn> <msub> <mi>M</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&amp;CenterDot;</mo> <msub> <mi>M</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> </mrow> <mrow> <msubsup> <mi>M</mi> <mn>1</mn> <mn>2</mn> </msubsup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msubsup> <mi>M</mi> <mn>2</mn> <mn>2</mn> </msubsup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> </mrow> </mfrac> </mrow>

<mrow> <msub> <mi>S</mi> <mrow> <mi>c</mi> <mi>n</mi> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <mn>2</mn> <msub> <mi>N</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&amp;CenterDot;</mo> <msub> <mi>N</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> </mrow> <mrow> <msubsup> <mi>N</mi> <mn>1</mn> <mn>2</mn> </msubsup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msubsup> <mi>N</mi> <mn>2</mn> <mn>2</mn> </msubsup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>C</mi> <mn>1</mn> </msub> </mrow> </mfrac> </mrow>

S_c(x)＝S_cm(x)·S_cn(x)

wherein C is₁Is a normal number, M₁(x) And M₂(x) Respectively representing the gray value of the M chrominance channel of the reference image and the gray value of the M chrominance channel of the image to be evaluated, N₁(x) And N₂(x) Respectively representing the gray value of the M chroma channel of the reference image and the gray value of the N chroma channel of the image to be evaluated, S_cm(x) For M chroma channel similarity features, S_cn(x) Is an N chroma channel similarity feature.

4. The comprehensive evaluation method for the quality of the color images according to the claim 2 or 3, wherein the specific method for obtaining the similar characteristics of phase consistency by using log-Gabor wavelets in the step (3) comprises the following steps:

(31) calculating the position of a reference image and an image to be evaluated L brightness channel image point x in the direction of theta by using a two-dimensional log-Gabor filter_jEven symmetric filter response at sum scale nAnd odd symmetric filter responseThe two-dimensional log-Gabor filter expression is:

<mrow> <mi>G</mi> <mrow> <mo>(</mo> <mi>&amp;omega;</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mo>=</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>lg</mi> <mo>(</mo> <mfrac> <mi>&amp;omega;</mi> <msub> <mi>&amp;omega;</mi> <mn>0</mn> </msub> </mfrac> <mo>)</mo> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msup> <mrow> <mo>(</mo> <mi>lg</mi> <mo>(</mo> <mfrac> <mi>k</mi> <msub> <mi>&amp;omega;</mi> <mn>0</mn> </msub> </mfrac> <mo>)</mo> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </mfrac> <mo>)</mo> </mrow> <mo>&amp;CenterDot;</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <mfrac> <msup> <mrow> <mo>(</mo> <mi>&amp;theta;</mi> <mo>-</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mrow> <mn>2</mn> <msubsup> <mi>&amp;sigma;</mi> <mi>&amp;theta;</mi> <mn>2</mn> </msubsup> </mrow> </mfrac> <mo>)</mo> </mrow> </mrow>

whereinJ denotes the jth direction, J denotes the number of directions, σ_θA bandwidth for determining a filter angle; θ represents the filter direction angle; ω represents the angular frequency of the filter; omega₀Representing the center frequency of a two-bit log-Gabor filter, k representing the shape parameter of the filter, the shape of the filter being defined by ω₀And k.

(32) Calculate θ in the direction_jSum amplitude with sum scale nAnd direction of theta_jIn response to local energy

<mrow> <msub> <mi>A</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <msqrt> <mrow> <msup> <msub> <mi>e</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </mrow> </msub> <mn>2</mn> </msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msup> <msub> <mi>o</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </mrow> </msub> <mn>2</mn> </msup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> </msqrt> </mrow>

<mrow> <msub> <mi>E</mi> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <msqrt> <mrow> <msup> <mrow> <mo>(</mo> <munder> <mo>&amp;Sigma;</mo> <mi>n</mi> </munder> <msub> <mi>e</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </mrow> </msub> <mo>(</mo> <mi>x</mi> <mo>)</mo> <mo>)</mo> </mrow> <mn>2</mn> </msup> <mo>+</mo> <msup> <mrow> <mo>(</mo> <munder> <mo>&amp;Sigma;</mo> <mi>n</mi> </munder> <msub> <mi>o</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </mrow> </msub> <mo>(</mo> <mi>x</mi> <mo>)</mo> <mo>)</mo> </mrow> <mn>2</mn> </msup> </mrow> </msqrt> </mrow>

(33) And (3) synthesizing the responses of all dimensions in all directions, and calculating the phase consistency of all points:

<mrow> <mi>P</mi> <mi>C</mi> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfrac> <mrow> <munder> <mo>&amp;Sigma;</mo> <mi>j</mi> </munder> <msub> <mi>E</mi> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> </mrow> <mrow> <munder> <mo>&amp;Sigma;</mo> <mi>j</mi> </munder> <munder> <mo>&amp;Sigma;</mo> <mi>n</mi> </munder> <msub> <mi>A</mi> <mrow> <mi>n</mi> <mo>,</mo> <msub> <mi>&amp;theta;</mi> <mi>j</mi> </msub> </mrow> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <mi>&amp;epsiv;</mi> </mrow> </mfrac> </mrow>

wherein ε is a normal number;

calculating the phase consistency characteristic PC of the brightness channel of the reference image by the same method₁(x) And the phase consistency characteristic PC of the brightness channel of the image to be evaluated₂(x)；

(34) Calculating similar characteristics of phase consistency of the reference image and the L brightness channel image of the image to be evaluated pixel by pixel:

<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> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>&amp;CenterDot;</mo> <msub> <mi>PC</mi> <mn>2</mn> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>C</mi> <mn>2</mn> </msub> </mrow> <mrow> <msubsup> <mi>PC</mi> <mn>1</mn> <mn>2</mn> </msubsup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msubsup> <mi>PC</mi> <mn>2</mn> <mn>2</mn> </msubsup> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>+</mo> <msub> <mi>C</mi> <mn>2</mn> </msub> </mrow> </mfrac> </mrow>

wherein, C₂Is a normal number.

5. The comprehensive evaluation method of color image quality according to claim 2 or 3, characterized in that the position saliency of each point in the step (4) is characterized by:

<mrow> <msub> <mi>S</mi> <mi>d</mi> </msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mo>=</mo> <mi>exp</mi> <mrow> <mo>(</mo> <mo>-</mo> <mfrac> <mrow> <mo>|</mo> <mo>|</mo> <mi>x</mi> <mo>-</mo> <msub> <mi>x</mi> <mi>c</mi> </msub> <mo>|</mo> <msubsup> <mo>|</mo> <mn>2</mn> <mn>2</mn> </msubsup> </mrow> <mrow> <msup> <msub> <mi>&amp;sigma;</mi> <mi>d</mi> </msub> <mn>2</mn> </msup> </mrow> </mfrac> <mo>)</mo> </mrow> </mrow>

wherein x is_cIs the feature map center point;representing the distance between the x point and the central point of the feature map; sigma_d ²Are empirical parameters.