A method and system for color image enhancement based on visual perception characteristics of individuals with red-green color vision abnormalities.

Abstract

This invention discloses a color image enhancement method and system for individuals with red-green color vision deficiencies, belonging to the field of image processing technology, and specifically relating to these methods based on visual perception characteristics. The enhancement method includes: determining the image enhancement type based on the type of red-green color vision deficiency; red-green color vision deficiency types include: green weakness, green blindness, severe red weakness, red blindness, and ordinary red weakness; image enhancement types include: a first enhancement type, a second enhancement type, and a third enhancement type; generating an abnormal cone spectral sensitivity curve based on the image enhancement type; measuring the spectral radiation distribution of the three primary colors of the image display device; obtaining a conversion matrix from RGB values ​​to LMS values ​​of the image based on the abnormal cone spectral sensitivity curve and the spectral radiation distribution of the three primary colors; and obtaining the RGB values ​​of the enhanced image based on the conversion matrix. This method eliminates the need for precise color weakness levels and allows for the use of a unified enhancement algorithm for both green blindness and green weakness individuals, thus achieving color image enhancement.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, and specifically relates to a method and system for color image enhancement for individuals with red-green color vision abnormalities based on their visual perception characteristics. Background Technology

[0002] People with red-green color vision deficiency have an impaired ability to extract information from images and videos due to a decline in their color discrimination ability. Enhancing their quality of life through color images and videos can significantly improve their condition. Approximately 75% of people with color vision deficiency are red-green color-weak, a proportion far greater than that of people with red-green color blindness.

[0003] Existing technologies require the design of different color image enhancement algorithms for red-green color blindness and red-green color weakness. Color image enhancement methods for red-green color blindness mainly include: custom difference summation, hue rotation, and optimization methods. Color image enhancement methods for red-green color weakness mainly include: analyzing the color perception differences between normal color visioners and those with red-green color weakness from chromaticity ratio and hue difference; establishing a color shift model in the chromaticity space to control the degree of color difference to achieve image enhancement; enhancing the color perception of those with red-green color weakness in the color space while preserving the naturalness of the enhanced image; and image enhancement based on the correlation between luminance and chromaticity information content. These methods primarily rely on the color blindness confusion line in the CIE chromaticity diagram as their theoretical basis and employ various algorithms in the field of image processing to achieve enhancement, without considering the visual perception characteristics of individuals with red-green color aberrations. Related research indicates that individuals with red-green color aberrations have unique characteristics in distinguishing the colors of objects in color images; their color discrimination ability is not as low as predicted by the color blindness confusion line due to the influence of the distribution of color information in the color image.

[0004] Related research indicates that red-green color weakness image enhancement methods based on the visual perception characteristics of individuals with abnormal color vision require designing enhanced images of varying degrees according to the severity of color weakness. This is achieved by simulating different levels of red and green weakness by shifting the spectral sensitivity curves of normal L or M cones to varying degrees, thereby enhancing the image. For example, existing methods, based on the principle that the severity of the FM100 test is proportional to its severity, generate enhanced images according to the observer's FM100 test score, dividing the color weakness into 10 levels and shifting the cone spectral sensitivity curve at 2nm intervals to obtain the abnormal cone spectral sensitivity curve.

[0005] Existing methods have the following drawbacks: First, completely different image enhancement methods are used for red-green color blindness and color weakness. This could lead to significant color changes in the enhanced image for red-green color blindness, resulting in a very unpleasant subjective experience for those with red-green color weakness. Second, different levels of enhancement are needed depending on the severity of color weakness. However, accurately determining the degree of color weakness for each individual is difficult in practice, requiring specialized equipment. Furthermore, the enhancement effect of color images cannot be determined when the accuracy of the severity of color weakness is unknown.

[0006] Therefore, there is an urgent need for a color image enhancement method and system for individuals with red-green color vision abnormalities based on their visual perception characteristics, in order to address the aforementioned shortcomings of existing methods. Summary of the Invention

[0007] The purpose of this invention is to provide a color image enhancement method for individuals with red-green color vision abnormalities based on their visual perception characteristics. The method is characterized by the following steps:

[0008] Step S1: Determine the image enhancement type based on the red-green color vision abnormality type; the red-green color vision abnormality type includes: green weakness, green blindness, severe red weakness, red blindness type, and ordinary red weakness; the image enhancement type includes: first enhancement type, second enhancement type, and third enhancement type;

[0009] Step S2: Based on the image enhancement type, generate an abnormal cone spectral sensitivity curve;

[0010] Step S3: Measure the spectral radiation distribution of the three primary colors of the image display device;

[0011] Step S4: Obtain the conversion matrix from RGB values ​​to LMS values ​​of the image based on the abnormal cone spectral sensitivity curve and the three primary color spectral radiation distribution;

[0012] Step S5: Obtain the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement.

[0013] The step S1 of determining the image enhancement type based on the red-green color vision abnormality type includes: if the red-green color vision abnormality type is green weakness or green blindness, the image enhancement type is the first enhancement type; if the red-green color vision abnormality type is severe red weakness or red blindness, the image enhancement type is the second enhancement type; if the red-green color vision abnormality type is ordinary red weakness, the image enhancement type is the third enhancement type.

[0014] The specific steps for generating the abnormal cone spectral sensitivity curve in step S2 include:

[0015] Step S21: Based on the image enhancement type, obtain the cone spectral sensitivity curve of a normal color vision individual;

[0016] Step S22: Obtain the scaling function based on the image enhancement type;

[0017] Step S23: Obtain the abnormal cone spectral sensitivity curve based on the cone spectral sensitivity curve and the scaling function of a normal color visioner.

[0018] If the image enhancement type is: First enhancement type:

[0019] The spectral sensitivity curve of a normal color visioner is: M-cone spectral sensitivity function M of a normal color visioner normal (λ);

[0020] The proportional function is: the first proportional function b(λ);

[0021] The abnormal cone spectral sensitivity curve is: Abnormal M-cone spectral sensitivity function M d (λ);

[0022] M d (λ)=M normal (λ) / b(λ) (1.1)

[0023] If the image enhancement type is: Second enhancement type:

[0024] The spectral sensitivity curve of a normal color visioner is: L-cone spectral sensitivity function of a normal color visioner L normal (λ);

[0025] The proportional function is: the second proportional function c(λ);

[0026] The abnormal cone spectral sensitivity curve is: Severely abnormal L cone spectral sensitivity function L p (λ);

[0027] L p (λ)=L normal (λ) / c(λ) (1.2)

[0028] If the image enhancement type is: Third enhancement type:

[0029] The spectral sensitivity curve of a normal color visioner is: L-cone spectral sensitivity function of a normal color visioner L normal (λ);

[0030] The proportional function is: the third proportional function d(λ);

[0031] The abnormal cone spectral sensitivity curve is: General abnormal L-cone spectral sensitivity function L pa (λ);

[0032] L pa (λ)=L normal(λ) / d(λ)(1.3).

[0033] The transformation matrix Γ from RGB values ​​to LMS values ​​in step S4 is:

[0034]

[0035] The constraints on the transformation matrix Γ are:

[0036]

[0037] Where, L R L is the L-cone response value corresponding to the red primary color of the image display device. G L is the L-cone response value corresponding to the green primary color of the image display device. B M is the L-cone response value corresponding to the blue primary color of the image display device. R M is the M-cone response value corresponding to the red primary color of the image display device. G M is the M-cone response value corresponding to the green primary color of the image display device. B S represents the M-cone response value corresponding to the blue primary color of the image display device. R S is the S-cone response value corresponding to the red primary color of the image display device. G S is the S-cone response value corresponding to the green primary color of the image display device. B The S-cone response value corresponding to the blue primary color of the image display device; ρ L ρ M and ρ S The normalization coefficients ensure that all different levels of gray for both normal and abnormal color visioners have the same value in the RGB space and the independent color space of various devices within the range of (0, 0, 0) to (1, 1, 1); L(λ) is the L-cone spectral sensitivity function, M(λ) is the M-cone spectral sensitivity function, and S(λ) is the S-cone spectral sensitivity function.

[0038] The step S5, which involves obtaining the RGB values ​​of the enhanced image based on the transformation matrix from RGB to LMS values, specifically includes:

[0039] Step S51: Input the RGB values ​​of the original image and perform gamma correction to obtain the corrected RGB values;

[0040] Step S52: Transformation matrix Г based on the RGB values ​​to LMS values ​​corresponding to normal color vision individuals normal The corrected RGB values ​​are converted to LMS values;

[0041] Step S53: Based on the image enhancement type, select a color vision anomaly transformation matrix, convert the LMS value into an enhanced RGB value based on the color vision anomaly transformation matrix, and perform gamma operation on the enhanced RGB value to obtain the RGB value of the enhanced image.

[0042] The selection of the color vision anomaly transformation matrix based on the image enhancement type includes:

[0043] If the image enhancement type is: First enhancement type:

[0044] The color vision deficiency conversion matrix is: the conversion matrix Г of RGB values ​​to LMS values ​​for individuals with green blindness or green weakness. deutan ;

[0045] If the image enhancement type is: Second enhancement type:

[0046] The color vision deficiency conversion matrix is: the conversion matrix Г of RGB values ​​to LMS values ​​for individuals with severe red weakness or red blindness. protan ;

[0047] If the image enhancement type is: Third enhancement type:

[0048] The color vision deficiency transformation matrix is: the transformation matrix Г for the RGB values ​​to LMS values ​​corresponding to ordinary red-weak individuals. protanomaly .

[0049] Another object of the present invention is to provide an image enhancement system for a color image enhancement method for red-green color vision abnormalities based on visual perception characteristics, as described in the present invention, characterized in that it comprises:

[0050] Image enhancement type determination module: Determines the image enhancement type based on the red-green color vision abnormality type; the red-green color vision abnormality type includes: green weakness, green blindness, severe red weakness, red blindness type, and ordinary red weakness; the image enhancement type includes: first enhancement type, second enhancement type, and third enhancement type;

[0051] Abnormal cone spectral sensitivity curve generation module: Generates abnormal cone spectral sensitivity curves based on the image enhancement type;

[0052] The three-primary-color spectral radiation distribution measurement module measures the three-primary-color spectral radiation distribution of image display devices.

[0053] Image RGB to LMS value conversion matrix acquisition module: Based on the anomalous cone spectral sensitivity curve and the three primary color spectral radiation distribution, the image RGB to LMS value conversion matrix is ​​obtained;

[0054] Enhanced Image RGB Value Acquisition Module: Obtains the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement.

[0055] The image enhancement type determination module determines the image enhancement type based on red-green color vision anomaly type, including:

[0056] If the red-green color vision deficiency type is green weakness or green blindness, the image enhancement type is the first enhancement type; if the red-green color vision deficiency type is severe red weakness or red blindness, the image enhancement type is the second enhancement type; if the red-green color vision deficiency type is ordinary red weakness, the image enhancement type is the third enhancement type.

[0057] The beneficial effects of this invention are as follows:

[0058] This invention discloses a color image enhancement method for individuals with red-green color vision abnormalities based on visual perception characteristics. The method involves: determining the image enhancement type based on the type of red-green color vision abnormality; generating an abnormal cone spectral sensitivity curve based on the image enhancement type; measuring the spectral radiation distribution of the three primary colors of the image display device; obtaining a conversion matrix from RGB values ​​to LMS values ​​of the image based on the abnormal cone spectral sensitivity curve and the spectral radiation distribution of the three primary colors; and obtaining the RGB values ​​of the enhanced image based on the conversion matrix, thus achieving color image enhancement.

[0059] This invention overcomes the problem in existing technologies where specific image enhancement algorithms are used for red-green color blindness and color weakness, resulting in significant color changes in the enhanced images for those with red-green color blindness, leading to poor subjective perception for them. The enhancement method disclosed in this invention eliminates the need for separate image enhancement methods for green-green color blindness and color weakness; both individuals with green-green color blindness and color weakness perceive the enhancement effect as good.

[0060] This invention overcomes the problems of existing technologies, which require obtaining images with different levels of enhancement based on the severity of color weakness, necessitate specialized equipment, and cannot determine the enhancement effect when the accuracy of the color weakness severity is unknown. The enhancement method disclosed in this invention provides images with good enhancement effects for different degrees of color weakness, eliminates the need for front-end measurement of color weakness severity, and avoids different enhancement levels based on severity. It allows for a single enhanced image to be used for individuals with different degrees of color weakness, and the enhancement effect is acceptable to them. Attached Figure Description

[0061] Figure 1 This is a flowchart illustrating the color image enhancement method for red-green color vision abnormalities based on visual perception characteristics according to the present invention.

[0062] Figure 2 This is a schematic diagram of the workflow of the image enhancement system according to an embodiment of the present invention;

[0063] Figure 3This is a graphical schematic diagram of the first proportional function b(λ) according to an embodiment of the present invention;

[0064] Figure 4 This is a graphical schematic diagram of the second proportional function c(λ) according to an embodiment of the present invention;

[0065] Figure 5 This is a graphical schematic diagram of the third proportional function d(λ) according to an embodiment of the present invention;

[0066] Figure 6 This is a schematic diagram of the enhanced image generated by the color image enhancement method for red-green color vision abnormalities based on visual perception characteristics, according to an embodiment of the present invention. Detailed Implementation

[0067] This invention provides a method and system for color image enhancement for individuals with red-green color vision abnormalities based on their visual perception characteristics. The invention will be further described in detail below with reference to the accompanying drawings.

[0068] like Figure 1 The embodiment of the present invention disclosed herein discloses a color image enhancement method for individuals with red-green color vision abnormalities based on their visual perception characteristics, comprising:

[0069] Step S1: Determine the image enhancement type based on the red-green color vision abnormality type; the red-green color vision abnormality type includes: green weakness, green blindness, severe red weakness, red blindness type, and ordinary red weakness; the image enhancement type includes: first enhancement type, second enhancement type, and third enhancement type;

[0070] Step S2: Based on the image enhancement type, generate an abnormal cone spectral sensitivity curve;

[0071] Step S3: Measure the spectral radiation distribution of the three primary colors of the image display device;

[0072] Step S4: Obtain the conversion matrix from RGB values ​​to LMS values ​​of the image based on the abnormal cone spectral sensitivity curve and the three primary color spectral radiation distribution;

[0073] Step S5: Obtain the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement.

[0074] In this embodiment, individuals with red-green color vision abnormalities are divided into three types: green blindness and green weakness, normal red weakness, and severe red weakness and red blindness. For each type, an image enhancement method is designed to maximize the enhancement effect for all individuals within each type. Unlike existing technologies that require specialized measurement equipment and methods, such as the FM100 test, and generate enhanced images based on the observer's FM100 test score, this invention discloses a color image enhancement method for red-green color vision abnormalities based on visual perception characteristics. This method obtains the spectral sensitivity function of abnormal L or M cones by dividing the spectral sensitivity function of normal L or M cones by a coefficient function. Then, it calculates the conversion matrix from RGB values ​​to LMS values ​​for the abnormal color vision abnormalities based on the product of the spectral sensitivity function and the three primary color radiation spectrum of the terminal display device, thereby achieving color image enhancement. The present invention does not require high accuracy of the front-end red-green color vision abnormality detection system. That is, it is sufficient to divide the red-green color vision abnormality group into three types. It is not required to obtain the degree of color weakness, and the corresponding images can be obtained in the opinion of the individuals who believe that the enhancement effect is good.

[0075] The specific implementation process for each step is described below.

[0076] Step S1: Determine the image enhancement type based on the red-green color vision abnormality type; the red-green color vision abnormality type includes: green weakness, green blindness, severe red weakness, red blindness type, and ordinary red weakness; the image enhancement type includes: first enhancement type, second enhancement type, and third enhancement type;

[0077] The step S1, which determines the image enhancement type based on the red-green color vision anomaly type, includes:

[0078] If the red-green vision deficiency type is green weakness or green blindness, then the image enhancement type is the first enhancement type;

[0079] If the red-green vision abnormality type is severe red weakness or red blindness, then the image enhancement type is the second enhancement type;

[0080] If the red-green color vision abnormality type is ordinary red weakness, then the image enhancement type is the third enhancement type;

[0081] In this embodiment, the accuracy requirement for the front-end red-green color vision deficiency detection system is not high. It is not necessary to obtain a specific degree of color weakness; it is sufficient to classify the red-green color vision deficiency group into one of three enhancement types: a first enhancement type, a second enhancement type, or a third enhancement type. Those skilled in the art should be familiar with existing detection methods, such as the Yu Ziping version of the color blindness test chart, the Farnsworth D-15, the Farnsworth-Munsell 100-Hue, and the Neitz color blindness test (OT-II). In this embodiment, no specific limitation is made to the specific detection method. This avoids the difficulty in accurately obtaining the degree of color weakness in individuals with color vision deficiency during practical applications.

[0082] Step S2: Based on the image enhancement type, generate an abnormal cone spectral sensitivity curve;

[0083] The specific steps for generating the abnormal cone spectral sensitivity curve in step S2 include:

[0084] Step S21: Based on the image enhancement type, obtain the cone spectral sensitivity curve of a normal color vision individual;

[0085] Step S22: Obtain the scaling function based on the image enhancement type;

[0086] Step S23: Obtain the abnormal cone spectral sensitivity curve based on the cone spectral sensitivity curve and the scaling function of a normal color visioner.

[0087] In this embodiment, the pre-set L, M, and S cone spectral sensitivity functions for normal color vision individuals are respectively represented by L... normal (λ), M normal (λ), S normal (λ) indicates that, as those skilled in the art should know, the preset L, M, and S cone spectral sensitivity functions for a normal color visioner are values ​​defined by Stockman and Sharpe (2000) or by Smith and Pokorny (1975). In this embodiment, the spectral sensitivity curve for a normal color visioner is L... normal (λ), M normal Choose one from (λ).

[0088] If the image enhancement type is: First enhancement type:

[0089] The spectral sensitivity curve of a normal color visioner is: M-cone spectral sensitivity function M of a normal color visioner normal (λ);

[0090] The proportional function is: the first proportional function b(λ);

[0091] The abnormal cone spectral sensitivity curve is: Abnormal M-cone spectral sensitivity function M d (λ);

[0092] M d (λ)=M normal (λ) / b(λ) (1.1)

[0093] In this embodiment, the preset M-cone spectral sensitivity function M for normal color vision is... normal (λ) is: a value defined by Stockman and Sharpe (2000), or a value defined by Smith and Pokorny (1975);

[0094] In this embodiment, the graph of the first proportional function b(λ) is as follows: Figure 3 As shown, this value was obtained by analyzing experimental data obtained from subjective evaluation experiments of color images with different enhancement levels by red-green color vision abnormalities. The data of the first proportional function b(λ) are shown in Table 1. The enhanced color image derived from the abnormal M-cone spectral sensitivity function corresponding to this value has a good enhancement effect on both green blindness and green weakness.

[0095]

[0096] If the image enhancement type is: Second enhancement type:

[0097] The spectral sensitivity curve of a normal color visioner is: L-cone spectral sensitivity function of a normal color visioner L normal (λ);

[0098] The proportional function is: the second proportional function c(λ);

[0099] The abnormal cone spectral sensitivity curve is: Severely abnormal L cone spectral sensitivity function L p (λ);

[0100] L p (λ)=L normal (λ) / c(λ) (1.2)

[0101] In this embodiment, the preset L-cone spectral sensitivity function L for normal color vision is... normal (λ) is: a value defined by Stockman and Sharpe (2000), or a value defined by Smith and Pokorny (1975);

[0102] In this embodiment, the graph of the second proportional function c(λ) is as follows: Figure 4As shown, this value was obtained by analyzing experimental data obtained from subjective evaluation experiments of color images with different enhancement levels by red-green color vision abnormalities. The data of the second proportional function c(λ) are shown in Table 2. The enhanced color image derived from the abnormal L-cone spectral sensitivity function corresponding to this value has a good enhancement effect on severe red weakness and red blindness.

[0103]

[0104] If the image enhancement type is: Third enhancement type:

[0105] The spectral sensitivity curve of a normal color visioner is: L-cone spectral sensitivity function of a normal color visioner L normal (λ);

[0106] The proportional function is: the third proportional function d(λ);

[0107] The abnormal cone spectral sensitivity curve is: General abnormal L-cone spectral sensitivity function L pa (λ);

[0108] L pa (λ)=L normal (λ) / d(λ) (1.3)

[0109] In this embodiment, the graph of the third proportional function d(λ) is as follows: Figure 5 As shown, this value was obtained by analyzing experimental data obtained from subjective evaluation experiments of color images with different enhancement levels by individuals with red-green color vision abnormalities. The data of the third proportional function d(λ) is shown in Table 3. The enhanced color image derived from the abnormal L-cone spectral sensitivity function corresponding to this value has a good enhancement effect on ordinary red-weak individuals.

[0110]

[0111] Step S3: Measure the spectral radiance curve of the three primary colors of the image display device;

[0112] In this embodiment, the spectral radiance curves of the three primary colors (red, green, and blue) of the image display device are measured using specialized optical equipment. The spectral radiance curves of the three primary colors include: the spectral radiance curve of the red primary color. Spectral radiance curve of green base color and the spectral radiance curve of blue primary color

[0113] Step S4: Obtain the conversion matrix from RGB values ​​to LMS values ​​of the image based on the abnormal cone spectral sensitivity curve and the three primary color spectral radiation distribution;

[0114] The transformation matrix Γ from RGB values ​​to LMS values ​​in step S4 is:

[0115]

[0116] The constraints on the transformation matrix Γ are:

[0117]

[0118] Where, L R L is the L-cone response value corresponding to the red primary color of the image display device. G L is the L-cone response value corresponding to the green primary color of the image display device. B M is the L-cone response value corresponding to the blue primary color of the image display device. R M is the M-cone response value corresponding to the red primary color of the image display device. G M is the M-cone response value corresponding to the green primary color of the image display device. B S represents the M-cone response value corresponding to the blue primary color of the image display device. R S is the S-cone response value corresponding to the red primary color of the image display device. G S is the S-cone response value corresponding to the green primary color of the image display device. B The S-cone response value corresponding to the blue primary color of the image display device; ρ L ρ M and ρ S The normalization coefficients ensure that all different levels of gray for normal and abnormal color visioners, i.e., gray has the same value in the RGB space and the independent color space of various devices in the range of (0, 0, 0) to (1, 1, 1); L(λ) is the L-cone spectral sensitivity function, M(λ) is the M-cone spectral sensitivity function, and S(λ) is the S-cone spectral sensitivity function.

[0119] ρ L ρ M and ρ S The normalization coefficient is used to satisfy the constraints of the transformation matrix Γ defined in formula (3).

[0120] When the L-cone spectral sensitivity function L(λ) is the L-cone spectral sensitivity function of a normal color visioner, L... normal (λ); and the M-cone spectral sensitivity function M(λ) is the M-cone spectral sensitivity function M of a normal color visioner. normal (λ); and the S-cone spectral sensitivity function S(λ) is the S-cone spectral sensitivity function S of a normal color visioner. normal When (λ), the conversion matrix Г from RGB values ​​to LMS values ​​corresponding to normal color vision is obtained through equations (2), (3), and (4). normal ;

[0121] When the L-cone spectral sensitivity function L(λ) is the L-cone spectral sensitivity function of a normal color visioner, L... normal (λ); and the M-cone spectral sensitivity function M(λ) is an abnormal M-cone spectral sensitivity function M d (λ); and the S-cone spectral sensitivity function S(λ) is the S-cone spectral sensitivity function S of a normal color visioner. normal When (λ), the conversion matrix Г from RGB values ​​to LMS values ​​corresponding to green-blind or green-weak individuals is obtained through equations (2), (3), and (4). deutan ;

[0122] When the L-cone spectral sensitivity function L(λ) is severely abnormal, the L-cone spectral sensitivity function L p (λ); and the M-cone spectral sensitivity function M(λ) is the M-cone spectral sensitivity function M of a normal color visioner. normal (λ); and the S-cone spectral sensitivity function S(λ) is the S-cone spectral sensitivity function S of a normal color visioner. normal When (λ), the conversion matrix Г from RGB values ​​to LMS values ​​for individuals with severe red weakness or red blindness is obtained through equations (2), (3), and (4). protan ;

[0123] When the L-cone spectral sensitivity function L(λ) is a general anomalous L-cone spectral sensitivity function L pa (λ); and the M-cone spectral sensitivity function M(λ) is the M-cone spectral sensitivity function M of a normal color visioner. normal (λ); and the S-cone spectral sensitivity function S(λ) is the S-cone spectral sensitivity function S of a normal color visioner. normal When (λ), the conversion matrix Г from RGB value to LMS value corresponding to the ordinary red weak element is obtained through equations (2), (3) and (4). protanomaly ;

[0124] Step S5: Obtain the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement.

[0125] The step S5, which involves obtaining the RGB values ​​of the enhanced image based on the transformation matrix from RGB to LMS values, specifically includes:

[0126] Step S51: Input the RGB values ​​of the original image and perform gamma correction to obtain the corrected RGB values;

[0127] Step S52: Transformation matrix Г based on the RGB values ​​to LMS values ​​corresponding to normal color vision individuals normal The corrected RGB values ​​are converted to LMS values;

[0128] Step S53: Based on the image enhancement type, select a color vision anomaly transformation matrix, convert the LMS value into an enhanced RGB value based on the color vision anomaly transformation matrix, and perform gamma operation on the enhanced RGB value to obtain the RGB value of the enhanced image.

[0129] Based on the image enhancement type, the color vision anomaly transformation matrix is ​​selected as follows:

[0130] If the image enhancement type is: First enhancement type:

[0131] The color vision deficiency conversion matrix is: the conversion matrix Г of RGB values ​​to LMS values ​​for individuals with green blindness or green weakness. deutan ;

[0132] If the image enhancement type is: Second enhancement type:

[0133] The color vision deficiency conversion matrix is: the conversion matrix Г of RGB values ​​to LMS values ​​for individuals with severe red weakness or red blindness. protan ;

[0134] If the image enhancement type is: Third enhancement type:

[0135] The color vision deficiency transformation matrix is: the transformation matrix Г for the RGB values ​​to LMS values ​​corresponding to ordinary red-weak individuals. protanomaly .

[0136] Another embodiment of the present invention discloses an image enhancement system for a color image enhancement method for red-green color vision aberrant individuals based on visual perception characteristics, comprising:

[0137] Image enhancement type determination module: Determines the image enhancement type based on the red-green color vision abnormality type; the red-green color vision abnormality type includes: green weakness, green blindness, severe red weakness, red blindness type, and ordinary red weakness; the image enhancement type includes: first enhancement type, second enhancement type, and third enhancement type;

[0138] Abnormal cone spectral sensitivity curve generation module: Generates abnormal cone spectral sensitivity curves based on the image enhancement type;

[0139] The three-primary-color spectral radiation distribution measurement module measures the three-primary-color spectral radiation distribution of image display devices.

[0140] Image RGB to LMS value conversion matrix acquisition module: Based on the anomalous cone spectral sensitivity curve and the three primary color spectral radiation distribution, the image RGB to LMS value conversion matrix is ​​obtained;

[0141] Enhanced Image RGB Value Acquisition Module: Obtains the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement.

[0142] The image enhancement type determination module determines the image enhancement type based on red-green color vision anomaly type, including:

[0143] If the red-green color vision deficiency type is green weakness or green blindness, the image enhancement type is the first enhancement type; if the red-green color vision deficiency type is severe red weakness or red blindness, the image enhancement type is the second enhancement type; if the red-green color vision deficiency type is ordinary red weakness, the image enhancement type is the third enhancement type.

[0144] In this embodiment, the workflow of the image enhancement system of the color image enhancement method for red-green color vision abnormalities based on visual perception characteristics according to the present invention is as follows: Figure 2 As shown, the front-end color blindness and color weakness detection system only needs to determine the type of red-green color vision abnormality in individuals with color blindness or color weakness. These abnormality types include: green weakness, green blindness, severe red weakness, red blindness, and ordinary red weakness; there is no need to differentiate the specific severity of color blindness. The image enhancement type determination module determines the image enhancement type based on the red-green color vision abnormality type; these types include: a first enhancement type, a second enhancement type, and a third enhancement type.

[0145] With the coordinated efforts of the image enhancement type determination module, the abnormal cone spectral sensitivity curve generation module, the three primary color spectral radiation distribution measurement module, the image RGB value to LMS value conversion matrix acquisition module, and the enhanced image RGB value acquisition module, three processing links are formed corresponding to the three image enhancement types. Figure 2 Image enhancement module 1, image enhancement module 2 and image enhancement module 3.

[0146] If the red-green vision abnormality type is green weakness or green blindness, then the image enhancement type is the first enhancement type, and the image enhancement module 1 link is executed;

[0147] If the red-green vision abnormality type is severe red weakness or red blindness, then the image enhancement type is the second enhancement type, and the image enhancement module 2 link is executed;

[0148] If the red-green color vision abnormality type is ordinary red weakness, then the image enhancement type is the third enhancement type, and the image enhancement module 3 link is executed.

[0149] To verify the effectiveness of the color image enhancement method and system for red-green color vision abnormalities based on visual perception characteristics disclosed in this invention, the following verification experiment was conducted.

[0150] A total of 15 individuals with red-green color vision deficiency and 4 individuals with normal color vision participated in the experiment, aged between 20 and 25. The 15 individuals with red-green color vision deficiency underwent the following four tests: Yu Ziping version of the color blindness test chart, Farnsworth D-15, Farnsworth-Munsell 100-Hue and Neitz color blindness test (OT-II), and were classified as 5 with green weakness, 4 with red weakness, 4 with green blindness, and 2 with red blindness.

[0151] The experiment used 23 original images as stimuli. The scenes in the images contained a variety of colors as much as possible, such as... Figure 5 As shown. These images are from the Kodak test image library, sRGB color images corresponding to multispectral images from the Columbia University Color Vision Laboratory, and sRGB color images corresponding to multispectral images from the Munsell Color Science Laboratory. In each experiment, the original image and the enhanced image were displayed side by side on a gray background with (R,G,B)=(128,128,128) and a brightness value of 16.5cd / m2.

[0152] The observer first acclimatized in a dark room for 3 minutes before entering the experiment. Each experiment consisted of 23 trials, corresponding to 23 images, presented in a random order. In each trial, the original image was on the left (or right), and the enhanced image was on the right (or left). The initial value of the enhanced image was the same as the original image, with no enhancement. The observer adjusted the enhancement level using a button. For each image, the enhanced image consisted of images generated by Yang's algorithm with different enhancement levels for weak red or weak green, and an image generated by the algorithm in this patent. The observer's task was to "adjust the enhancement level of the enhanced image to achieve the best enhancement effect compared to the original image on the left, i.e., the strongest image contrast and a very natural image." Contrast and naturalness are two of the most important factors in measuring the effect of image enhancement. Contrast can be understood as the difference within a local area or the difference between distant areas in an image, such as the color difference between objects. When the contrast of an image is increased, it can easily cause the colors to appear unnatural to those with weak red and green. Therefore, preserving naturalness is also an important factor in evaluating the effectiveness of an algorithm. Here, the observer's task was to select the enhanced image that simultaneously considered both contrast and naturalness.

[0153] Each observer repeated the experiment 6 times. For each observer, the image with the best enhancement effect was the average of the 6 experiments, and the best image enhancement level was the average of the 23 images.

[0154] Experimental results show that the images with the best enhancement effect obtained by each group of observers in the experiment were the enhanced images generated by the method in this patent. The images with the best enhancement effect for green-blindness and green-weakness were basically the same, meaning the enhancement levels were basically the same, and were the images generated by the algorithm in module 1 of this patent. The images with the best enhancement effect for red-weakness and red-blindness were different. The images with the best enhancement effect obtained by the three individuals with ordinary red-weakness were the images generated by module 3 of this patent, while the images with the best enhancement effect for the one individual with severe red-weakness and red-blindness were the images generated by module 2 of this patent. The images with the best enhancement effect selected by normal color visioners in the experiment were very close to those with green-blindness and green-weakness, indicating that image enhancement for individuals with green-blindness and green-weakness does not hinder the perception of images by normal color visioners. The experiment shows that there is no correlation between the total error score of the FM 100-Hue test for individuals with color vision deficiencies and the images with the best enhancement effect they selected; the degree of image enhancement cannot be directly predicted by the total error score of the FM 100-Hue test.

[0155] In summary, to overcome the following technical problems existing in the current technology for color image enhancement for individuals with red-green color vision abnormalities:

[0156] (1) Existing technologies require the design of corresponding enhancement algorithms for four conditions: red-green color blindness, red-weakness, and green-weakness. This results in significant color changes in the enhanced images for red-green color blindness, leading to a very poor subjective experience for those with red-green weakness. Relevant data shows that red-green weakness accounts for approximately 75% of all color vision deficiencies, further highlighting the impact of this problem on the majority of color-weak individuals.

[0157] (2) Existing technologies require different levels of color image enhancement based on the severity of red and green color weakness to achieve good enhancement results. This necessitates the use of specialized equipment to accurately determine the degree of color weakness for each individual with color vision deficiency before achieving a good enhancement effect. Considering the poor portability of such specialized equipment, which requires setup in professional laboratories, this significantly increases the difficulty of accurately determining the severity of color weakness in ordinary individuals with color vision deficiency. This further highlights the problem that the enhancement effect of color images cannot be determined when the accuracy of the severity of color weakness is unknown.

[0158] The color image enhancement method for red-green color vision aberrant individuals based on visual perception characteristics disclosed in this invention has the following advantages:

[0159] (1) Although different enhancement algorithms are still needed for red weakness and red blindness, a single enhancement algorithm can be used for green blindness and green weakness.

[0160] In this embodiment, for red-light vision deficiency, two types of image enhancement methods are still needed: one for ordinary red-light weakness and the other for severe red-light weakness and red-light blindness, the difference being the degree of enhancement. For green-light vision deficiency, i.e., green-light blindness and green-light weakness, only one type of image enhancement method is used. The advantage of this invention is that it does not require designing separate image enhancement methods for green-light blindness and color weakness; for the enhanced image, both green-light blindness and color weakness are considered to have good enhancement effects.

[0161] (2) The algorithm obtains an enhanced image that has a good enhancement effect on different degrees of color weakness. It does not require the front end to measure the severity of color weakness, nor does it require different degrees of enhancement based on the severity of color weakness. Instead, all color weaknesses share the same enhanced image, but the enhancement effect is still acceptable.

[0162] Enhanced images generated using the color image enhancement method for red-green color vision abnormalities based on visual perception characteristics disclosed in this invention are as follows: Figure 6 As shown. In this embodiment, the accuracy requirement for the front-end red-green color vision abnormality detection system is not high. It is sufficient to distinguish between red-blindness, severe red-weakness, normal red-weakness, green-weakness, and green-blindness. It is not required to obtain the degree of color weakness level to obtain an image that is considered to have a good enhancement effect by all individuals. It can provide a default value for the color weakness color image enhancement system in the terminal display device. The enhanced image generated based on the default value is considered to have the best enhancement effect by many individuals with abnormal color vision. It is not necessary to generate an enhanced image based on the accurate severity of the user's individual color weakness; it directly obtains an enhanced image that is recognized by all individuals of this type of abnormal color vision and has a good enhancement effect.

Claims

1. A color image enhancement method for individuals with red-green color vision abnormalities based on their visual perception characteristics, characterized in that, Includes the following steps: Step S1: Determine the image enhancement type based on the red-green color vision anomaly type; The red-green color vision deficiency types include: green weakness, green blindness, severe red weakness, red blindness, and ordinary red weakness; the image enhancement types include: first enhancement type, second enhancement type, and third enhancement type; Step S2: Based on the image enhancement type, generate an abnormal cone spectral sensitivity curve; Step S3: Measure the spectral radiation distribution of the three primary colors of the image display device; Step S4: Obtain the conversion matrix from RGB values ​​to LMS values ​​of the image based on the abnormal cone spectral sensitivity curve and the three primary color spectral radiation distribution; Step S5: Obtain the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement; The step S1 of determining the image enhancement type based on the red-green color vision abnormality type includes: if the red-green color vision abnormality type is green weakness or green blindness, the image enhancement type is the first enhancement type; if the red-green color vision abnormality type is severe red weakness or red blindness, the image enhancement type is the second enhancement type; if the red-green color vision abnormality type is ordinary red weakness, the image enhancement type is the third enhancement type. The specific steps for generating the abnormal cone spectral sensitivity curve in step S2 include: Step S21: Based on the image enhancement type, obtain the cone spectral sensitivity curve of a normal color vision individual; Step S22: Obtain the scaling function based on the image enhancement type; Step S23: Obtain the abnormal cone spectral sensitivity curve based on the cone spectral sensitivity curve and the scaling function of a normal color visioner; The transformation matrix Γ from RGB values ​​to LMS values ​​in step S4 is: The constraints on the transformation matrix Γ are: Where, L R L is the L-cone response value corresponding to the red primary color of the image display device. G L is the L-cone response value corresponding to the green primary color of the image display device. B M is the L-cone response value corresponding to the blue primary color of the image display device. R M is the M-cone response value corresponding to the red primary color of the image display device. G M is the M-cone response value corresponding to the green primary color of the image display device. B S represents the M-cone response value corresponding to the blue primary color of the image display device. R S is the S-cone response value corresponding to the red primary color of the image display device. G S is the S-cone response value corresponding to the green primary color of the image display device. B The S-cone response value corresponding to the blue primary color of the image display device; ρ L ρ M and ρ S The normalization coefficients ensure that all different levels of gray, i.e., gray within the range of (0, 0, 0) to (1, 1, 1), have identical values ​​in both the RGB space and the independent color spaces of various devices, for both normal and abnormal color vision. L(λ) is the L-cone spectral sensitivity function, M(λ) is the M-cone spectral sensitivity function, and S(λ) is the S-cone spectral sensitivity function. The spectral radiance curves of the three primary colors include: the spectral radiance curve of the red primary color. Spectral radiance curve of green base color and the spectral radiance curve of blue primary color 2. The color image enhancement method for red-green color vision abnormalities based on visual perception characteristics according to claim 1, characterized in that, If the image enhancement type is: First enhancement type: The spectral sensitivity curve of a normal color visioner is: M-cone spectral sensitivity function M of a normal color visioner normal (λ); The proportional function is: the first proportional function b(λ); The abnormal cone spectral sensitivity curve is: Abnormal M-cone spectral sensitivity function M d (λ); M d (λ)=M normal (λ) / b(λ) (1.1) If the image enhancement type is: Second enhancement type: The spectral sensitivity curve of a normal color visioner is: L-cone spectral sensitivity function of a normal color visioner L normal (λ); The proportional function is: the second proportional function c(λ); The abnormal cone spectral sensitivity curve is: Severely abnormal L cone spectral sensitivity function L p (λ); L p (λ)=L normal (λ) / c(λ) (1.2) If the image enhancement type is: Third enhancement type: The spectral sensitivity curve of a normal color visioner is: L-cone spectral sensitivity function of a normal color visioner L normal (λ); The proportional function is: the third proportional function d(λ); The abnormal cone spectral sensitivity curve is: General abnormal L-cone spectral sensitivity function L pa (λ); L pa (λ)=L normal (λ) / d(λ) (1.3).

3. The color image enhancement method for red-green color vision abnormalities based on visual perception characteristics according to claim 1, characterized in that, The step S5, which involves obtaining the RGB values ​​of the enhanced image based on the transformation matrix from RGB to LMS values, specifically includes: Step S51: Input the RGB values ​​of the original image and perform gamma correction to obtain the corrected RGB values; Step S52: Transformation matrix Г based on the RGB values ​​to LMS values ​​corresponding to normal color vision individuals normal The corrected RGB values ​​are converted to LMS values; Step S53: Based on the image enhancement type, select a color vision anomaly transformation matrix, convert the LMS value into an enhanced RGB value based on the color vision anomaly transformation matrix, and perform gamma operation on the enhanced RGB value to obtain the RGB value of the enhanced image.

4. The color image enhancement method for red-green color vision abnormalities based on visual perception characteristics according to claim 3, characterized in that, The selection of the color vision anomaly transformation matrix based on the image enhancement type includes: If the image enhancement type is: First enhancement type: The color vision deficiency conversion matrix is: the conversion matrix Г of RGB values ​​to LMS values ​​for individuals with green blindness or green weakness. deutan ; If the image enhancement type is: Second enhancement type: The color vision deficiency conversion matrix is: the conversion matrix Г of RGB values ​​to LMS values ​​for individuals with severe red weakness or red blindness. protan ; If the image enhancement type is: Third enhancement type: The color vision deficiency transformation matrix is: the transformation matrix Г for the RGB values ​​to LMS values ​​corresponding to ordinary red-weak individuals. protanomaly .

5. An image enhancement system for a color image enhancement method for red-green color vision aberrant individuals based on visual perception characteristics according to claim 1, characterized in that, include: Image enhancement type determination module: Determines the image enhancement type based on red-green color vision anomaly type; The red-green color vision deficiency types include: green weakness, green blindness, severe red weakness, red blindness type, and ordinary red weakness; the image enhancement types include: first enhancement type, second enhancement type, and third enhancement type; Abnormal cone spectral sensitivity curve generation module: Generates abnormal cone spectral sensitivity curves based on the image enhancement type; The three-primary-color spectral radiation distribution measurement module measures the three-primary-color spectral radiation distribution of image display devices. Image RGB to LMS value conversion matrix acquisition module: Based on the anomalous cone spectral sensitivity curve and the three primary color spectral radiation distribution, the image RGB to LMS value conversion matrix is ​​obtained; Enhanced Image RGB Value Acquisition Module: Obtains the RGB values ​​of the enhanced image based on the transformation matrix from RGB values ​​to LMS values, thereby achieving color image enhancement; The image enhancement type determination module determines the image enhancement type based on red-green color vision anomaly type, including: If the red-green color vision deficiency type is green weakness or green blindness, then the image enhancement type is the first enhancement type; if the red-green color vision deficiency type is severe red weakness or red blindness, then the image enhancement type is the second enhancement type; if the red-green color vision deficiency type is ordinary red weakness, then the image enhancement type is the third enhancement type. The abnormal cone spectral sensitivity curve generation module, based on the image enhancement type, includes the following specific steps for generating the abnormal cone spectral sensitivity curve: Step S21: Based on the image enhancement type, obtain the cone spectral sensitivity curve of a normal color vision individual; Step S22: Obtain the scaling function based on the image enhancement type; Step S23: Obtain the abnormal cone spectral sensitivity curve based on the cone spectral sensitivity curve and the scaling function of a normal color visioner; The module for obtaining the conversion matrix from RGB values ​​to LMS values ​​of the image: Based on the abnormal cone spectral sensitivity curve and the three primary color spectral radiation distribution, the conversion matrix Γ from RGB values ​​to LMS values ​​of the image is: The constraints on the transformation matrix Γ are: Where, L R L is the L-cone response value corresponding to the red primary color of the image display device. G L is the L-cone response value corresponding to the green primary color of the image display device. B M is the L-cone response value corresponding to the blue primary color of the image display device. R M is the M-cone response value corresponding to the red primary color of the image display device. G M is the M-cone response value corresponding to the green primary color of the image display device. B S represents the M-cone response value corresponding to the blue primary color of the image display device. R S is the S-cone response value corresponding to the red primary color of the image display device. G S is the S-cone response value corresponding to the green primary color of the image display device. B The S-cone response value corresponding to the blue primary color of the image display device; ρ L ρ M and ρ S The normalization coefficients ensure that all different levels of gray, i.e., gray within the range of (0, 0, 0) to (1, 1, 1), have identical values ​​in both the RGB space and the independent color spaces of various devices, for both normal and abnormal color vision. L(λ) is the L-cone spectral sensitivity function, M(λ) is the M-cone spectral sensitivity function, and S(λ) is the S-cone spectral sensitivity function. The spectral radiance curves of the three primary colors include: the spectral radiance curve of the red primary color. Spectral radiance curve of green base color and the spectral radiance curve of blue primary color

Images