Gray scale gamma adjustment method and device based on fuzzy dissimilarity context intensity transformation and medium

Abstract

The invention discloses a gray scale gamma adjustment method and device based on fuzzy dissimilarity context intensity transformation and a medium, and relates to the technical field of displayers, the method comprises the steps that a fuzzy dissimilarity histogram is constructed based on an input image, and the histogram also provides an average difference value of each gray scale; performing truncation operation on the histogram to limit an excessive enhancement rate; in order to obtain better display fidelity in-situ quality, gamma correction is carried out; applying online text intensity transformation to reply natural features of the image so as to obtain a final enhanced image, and fusing the original image and the enhanced image; and further calculating the image saturation, and converting back to the RGB space based on the enhanced intensity and saturation. The method and the device are used for solving the problems of low efficiency and poor accuracy during gray-scale gamma adjustment in the prior art, and the gray-scale gamma adjustment can be quickly and accurately completed.

Description

Technical Field

[0001] This application relates to the field of display technology, and in particular to a grayscale gamma adjustment method, device and medium based on fuzzy anisotropy context intensity transformation. Background Technology

[0002] Self-emissive panels such as organic light-emitting diodes (OLEDs) are becoming increasingly popular due to their high color fidelity and fast response speed. However, this also leads to more complex adjustment curves. Traditional LUT methods are inefficient in adjusting large-scale grayscale points and cannot simultaneously achieve both brightness and chromaticity goals, resulting in poor adjustment accuracy.

[0003] Therefore, how to quickly and accurately adjust grayscale gamma is an important issue that the industry urgently needs to address. Summary of the Invention

[0004] In response to the aforementioned problems and technical requirements, the applicant proposes a grayscale gamma adjustment method, device, and medium based on fuzzy dissimilarity context intensity transformation, which solves the problems of low efficiency and poor accuracy in grayscale gamma adjustment in the prior art, and achieves fast and accurate grayscale gamma adjustment.

[0005] This application provides a grayscale gamma adjustment method based on fuzzy anisotropy context intensity transformation, the method comprising:

[0006] The acquired color image is converted from RGB space to HSI space, and a fuzzy dissimilarity histogram describing the global brightness contrast distribution is generated based on the brightness channel in HSI space.

[0007] Based on the image information of the fuzzy dissimilarity histogram, a clipping threshold is calculated, and new brightness is reallocated to pixels whose brightness is greater than the clipping threshold.

[0008] Calculate the gamma value corresponding to the pixel to be adjusted, and use the gamma value to correct the nonlinear response of the brightness;

[0009] Based on the nonlinear response, function values ​​are assigned to the smooth regions and texture detail regions of the image, and the brightness of the smooth regions and texture detail regions is fused using a preset intensity conversion formula and the function values.

[0010] The saturation of the fused image is calculated, and the image, which includes both saturation and brightness, is converted to the RGB space to obtain the final color image.

[0011] According to the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation provided in the embodiments of this application, a fuzzy dissimilarity histogram describing the global brightness contrast distribution is generated based on the brightness channels in the HSI space, including:

[0012] Based on the luminance channel, the following calculation process is performed for each pixel:

[0013] Based on pixel intensity, calculate the similarity value between the pixel and each pixel in the preset neighborhood; calculate the average of the similarity values ​​of all pixels in the preset neighborhood to obtain the similarity index of the pixel; obtain the complement of the similarity index to obtain the dissimilarity index, where pixel intensity is a numerical expression of brightness;

[0014] A fuzzy dissimilarity histogram is constructed based on the dissimilarity index, where the dissimilarity index is used to characterize the difference in intensity between a pixel and its neighboring pixels.

[0015] According to the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation provided in the embodiments of this application, the image information includes: the total number of pixels, brightness range, and maximum slope of the fuzzy dissimilarity histogram;

[0016] Calculating the clipping threshold based on the image information of the fuzzy dissimilarity histogram includes:

[0017] Input the total number of pixels, the brightness range, and the maximum slope into a preset shearing determination formula to obtain the shearing threshold output by the shearing determination formula;

[0018] The shear determination formula includes:

[0019]

[0020] Where α represents the clipping threshold; M represents the total number of pixels; N represents the brightness range; β represents the clipping factor; and S... m This indicates the maximum slope.

[0021] According to the grayscale gamma adjustment method based on fuzzy anisotropy context intensity transformation provided in the embodiments of this application, new brightness is reallocated to pixels whose brightness is greater than the clipping threshold, including:

[0022] The pixel intensity of the pixel to be adjusted is input into a preset mapping formula to obtain a new pixel intensity output by the mapping formula, where the pixel intensity is a numerical expression of brightness; the new pixel intensity is then assigned to the pixel to be adjusted.

[0023] The mapping formulas include:

[0024]

[0025] Among them, I c Let l represent the new pixel intensity, l represent the pixel intensity before assignment, and PDF(.) represent the probability density function. m Indicates the maximum pixel intensity level.

[0026] According to the grayscale gamma adjustment method based on fuzzy anisotropy context intensity transformation provided in the embodiments of this application, the gamma value corresponding to the pixel to be adjusted is calculated, including:

[0027] The pixel intensity of the adjusted pixel is input into a preset gamma parameter calculation formula to obtain the gamma value output by the gamma parameter calculation formula.

[0028] The formula for calculating gamma parameters includes:

[0029]

[0030] Where γ represents the gamma value and l represents the pixel intensity before allocation.

[0031] According to the grayscale gamma adjustment method based on fuzzy anisotropy context intensity transformation provided in the embodiments of this application, the nonlinear response of brightness is corrected using the gamma value, including:

[0032] The gamma value is input into a preset nonlinear description formula to obtain the nonlinear response output by the nonlinear description formula.

[0033] The nonlinear description formula includes:

[0034]

[0035] Among them, I g (x,y) represents the nonlinear response, I c This represents the new pixel intensity, where pixel intensity is a numerical expression of brightness, I. cmax γ represents the maximum new pixel intensity, and γ represents the gamma value.

[0036] According to the grayscale gamma adjustment method based on fuzzy anisotropy context intensity transformation provided in the embodiments of this application, the brightness of smooth regions and texture detail regions is fused using a preset intensity conversion formula and the function value, including:

[0037] The function value is input into the intensity conversion formula to obtain the brightness of the blended smooth area and texture detail area output by the intensity conversion formula;

[0038] The intensity conversion formula includes:

[0039] I e (x,y)=μc(x,y)I g(x,y)+μc(x,y)I(x,y);

[0040] Among them, I e (x,y) represents the enhanced strength after fusion, μc(x,y) represents the function value, and I g (x,y) represents the enhanced pixel intensity, and I(x,y) represents the original pixel intensity.

[0041] According to the grayscale gamma adjustment method based on fuzzy anisotropy context intensity transformation provided in the embodiments of this application, the saturation of the fused image is calculated, including:

[0042] Different saturation calculation formulas are used to calculate saturation for pixels of different intensities.

[0043] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation as described above.

[0044] This application also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation as described above.

[0045] The grayscale gamma adjustment method, device, and medium based on fuzzy dissimilarity context intensity transformation provided in this application first constructs a fuzzy dissimilarity histogram based on the input image, which also provides the average difference value for each grayscale level; then, the histogram is truncated to limit over-enhancement; gamma correction is performed to obtain better display fidelity; next, a context intensity transformation is applied to restore the natural features of the image, thereby obtaining the final enhanced image, and the original image and the enhanced image are fused; then, the image saturation is calculated, and the enhanced intensity and saturation are converted back to RGB space. This solves the problems of low efficiency and poor accuracy in grayscale gamma adjustment in the prior art, and achieves fast and accurate grayscale gamma adjustment. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] Figure 1This is one of the flowcharts of the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation provided in the embodiments of this application;

[0048] Figure 2 This is the second flowchart of the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation provided in the embodiments of this application;

[0049] Figure 3 This is a schematic diagram of brightness changes provided in an embodiment of this application;

[0050] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0052] To further clarify the background of this application:

[0053] Regarding developments in the field of technology:

[0054] With the continuous evolution of display technology, grayscale gamma adjustment, as a crucial element in improving image quality and visual experience, has undergone several transformations in its methods and tools. In the early CRT era, the brightness output of a monitor exhibited a significant non-linear relationship with the input voltage. This non-linearity was compensated for through hardware-level gamma correction circuits and manual adjustments to ensure natural transitions between light and dark areas in the image. With the advent of liquid crystal displays (LCDs) and light-emitting diodes (LEDs), digital control enabled gamma adjustment through lookup tables (LUTs), achieving greater precision, but still often relying on repeated manual experimentation and calibration.

[0055] With the widespread adoption of self-emissive panels such as organic light-emitting diodes (OLEDs), which offer high color fidelity and fast response speeds, the challenge of more complex adjustment curves has also arisen. Traditional LUT methods are inefficient in adjusting large-scale grayscale points and struggle to simultaneously achieve both brightness and chromaticity goals. To address this, researchers have introduced numerical optimization techniques such as mathematical fitting and least squares methods. By collecting data from key grayscale points and fitting curves, they have rapidly generated preliminary gamma curves.

[0056] In recent years, intelligent algorithms such as fuzzy control, genetic algorithms, and particle swarm optimization have been widely applied in the field of grayscale gamma adjustment. Fuzzy control can make flexible decisions based on the "grayscale" error between the target and reality, reducing reliance on precise models; global optimization techniques such as genetic algorithms and particle swarm optimization are better at finding optimal or near-optimal solutions in complex, multi-peaked objective function spaces. Some systems also combine machine learning, training models with historical calibration data to achieve automated, one-click rapid correction.

[0057] Meanwhile, the deep integration of the host computer software with the spectrophotometer and driver board enables online calibration during mass production. Colorimetric information is acquired in real time via serial port or network interface, and the calculation results are sent to the driver board; the entire process can be completed within minutes, greatly improving production efficiency. In the future, with the further development of neural networks and deep learning, gamma correction based on big data will become more intelligent and accurate, and may even be able to adapt and dynamically adjust to meet the personalized display effect requirements of different application scenarios.

[0058] Detailed explanation of existing technology:

[0059] In existing technologies, grayscale gamma adjustment mainly relies on two main methods: LUTs (Look-Up Tables) and manual adjustment. For traditional LCD and LED display panels, manufacturers typically embed a fixed-size gamma correction look-up table in the driver chip or panel controller. When the system receives the input grayscale value, it directly reads the corresponding output luminance or chromaticity value from the LUT using the address index and drives the backlight or pixel unit. This method is simple to implement and has a fast response, but its accuracy is limited. Due to the sparseness of the entries, linear or high-order interpolation must be performed on adjacent entries, which easily leads to quantization errors and stair-step effects in dark and bright areas. In addition, when calibration is required for different color spaces, a new LUT must be regenerated or downloaded, a process that is mostly offline and difficult to meet dynamic adjustment needs. If a fine adjustment of a certain grayscale range is required, the corresponding entries must be manually modified and reprogrammed, which is both time-consuming and unsuitable for large-scale production.

[0060] In early professional monitors and video processing equipment, maintenance personnel often manually adjusted parameters such as gamma, contrast, brightness, and color temperature through software interfaces to make the overall grayscale response closer to the target curve. The advantage of this method is its intuitive operation and ability to make fine adjustments for specific scenarios. However, its disadvantages are also quite obvious, such as: the adjustment results are highly dependent on the operator's experience and the observation environment, lacking repeatable quantitative indicators; each change requires immediate observation and repeated iterations, resulting in low efficiency and making it difficult to apply to large-scale production or real-time online calibration; and manual adjustments are mostly global gamma value settings, failing to account for the non-linear response differences in dark, mid, and highlight areas.

[0061] With the development of emerging panel technologies such as OLED and self-emissive Micro-LED, relying solely on LUTs and manual adjustments is no longer sufficient to meet the demands of modern displays for grayscale accuracy. Therefore, more and more research is introducing more sophisticated mathematical models and intelligent optimization algorithms. Curve fitting methods based on mathematical models first collect the actual brightness or color response of the panel at each input grayscale level, constructing a function model based on power-law functions, piecewise polynomials, or Chebyshev and Bézier curves. The optimal parameters are then solved using the least squares method to minimize the error between the theoretical and actual curves. These methods can quantify error distribution, automatically generate continuous curves, and effectively eliminate quantization errors generated in traditional LUT interpolation. However, the fitting process requires a large amount of measurement data and computing power, often resulting in a long offline calibration phase, making real-time deployment in resource-constrained embedded environments difficult.

[0062] At the intelligent optimization level, techniques such as fuzzy control, genetic algorithms, and particle swarm optimization (PSO) have also been introduced into grayscale adjustment. By designing fuzzy rules, the system can adjust the gamma value in real time according to the measurement error and its rate of change, achieving dynamic compensation for nonlinear drift of the panel. Genetic algorithms and PSO, on the other hand, encode grayscale correction parameters as chromosomes or particles and use the fitting error as the fitness, performing multiple iterations to globally search for the optimal solution. However, these algorithms require numerous iterations, and their convergence speed is limited by initial parameters and computing power; online real-time performance and resource consumption are pressing issues that need to be addressed.

[0063] In recent years, the application of automation and machine learning technologies has further promoted the development of grayscale gamma correction. Some solutions employ a closed-loop feedback mechanism that integrates photoelectric sensors on the panel to acquire the actual output brightness in real time, and then perform closed-loop compensation after subtracting it from the target curve, achieving millisecond-level dynamic correction. Other research utilizes convolutional neural networks (CNNs) to predict the optimal gamma adjustment scheme using multi-dimensional signals such as ambient light and temperature as input. Although these methods demonstrate strong adaptive capabilities, model training requires a large amount of calibration data, and deployment places high demands on hardware computing power and power consumption.

[0064] In summary, while existing technologies offer various approaches such as LUTs, manual adjustment, mathematical fitting, and intelligent optimization, they still face the challenge of balancing accuracy, efficiency, and cost. Future technological innovations can focus on achieving high-precision online curve fitting in resource-constrained environments. This could involve combining photoelectric feedback and machine learning to create an integrated "self-calibration + adaptive" gamma adjustment solution. Furthermore, dedicated calibration algorithms should be designed for different panels (such as the current-brightness nonlinearity of OLEDs and the backlight zoning effect of Mini-LEDs) to balance hardware cost and calibration accuracy.

[0065] Disadvantages of existing technology:

[0066] Mainstream display devices typically implement gamma correction through hardware lookup tables or registers, but this brings several limitations. First, the hardware cost is high: taking Mini-LED backlighting as an example, it requires thousands to tens of thousands of small LED zones and complex driving circuits. Although Mini-LED is cheaper than OLED, it still significantly increases hardware complexity and cost compared to traditional LCD. Furthermore, the hardware bit depth limits the accuracy of gamma adjustment. Many systems use 8- or 10-bit grayscale registers, looking up the output brightness through address indexes. This interpolation method is prone to quantization errors and makes it difficult to accurately match the ideal curve. Regarding response speed, some existing correction methods are slow, making it difficult to meet the rapid correction requirements of real-time or dynamic scenarios. Power consumption is also a significant issue: compared to OLED, LCD / mini-LED backlighting consumes less power at the same brightness. For example, a measurement showed that a 15.6-inch laptop screen at 200 nits brightness consumes a maximum of about 2.53W for an mLCD panel, while an OLED panel consumes 8.83W, meaning that LCD consumes only about one-third the power of OLED. Furthermore, techniques such as PWM dimming and frame rate control (FRC) used to increase bit depth may introduce flicker or frequency requirements, affecting reliability and power consumption. Therefore, current hardware implementations of gamma regulation have significant limitations in terms of cost, accuracy, power consumption, and response speed.

[0067] On the other hand, while intelligent algorithms such as fuzzy control, genetic algorithms, and particle swarm optimization introduced in recent years have improved the flexibility and automation of adjustment, these methods are generally computationally complex, time-consuming in iterative processes, and difficult to deploy efficiently in production lines. Furthermore, parameter tuning relies on expert experience, limiting their practicality. Simultaneously, existing methods generally lack consistent strategies for cross-device and cross-batch adjustment. Displays of different models or batches still need to be recalibrated individually before leaving the factory, increasing enterprise operation and maintenance and quality control costs. Gray-scale gamma correction algorithms are mostly based on power-law function fitting, but the nonlinear response of actual panels is more complex, and a single power law cannot completely cover all gray-scale regions. Traditional fitting accuracy is limited by the number of samples and measurement noise, easily leading to deviations in certain gray-scale levels. For example, existing technologies cannot determine a reasonable range when adjusting gray-scale gamma values, making rapid searching and accurate fitting difficult. In addition, RGB three-color channel correction is often performed separately; simply applying uniform correction to grayscale may ignore the chromaticity differences between channels. Moreover, after completing RGB gamma adjustment, it is often impossible to quickly correct the chromaticity coordinates, resulting in color reproduction deviations. Regarding low grayscale details, the algorithm's shortcomings are particularly evident: due to the drastic changes in brightness response at low grayscale levels, any error or quantization can cause "black compression" (loss of detail in dark areas). Studies have found that OLED displays exhibit significantly higher brightness at grayscale levels 3–8 than expected, leading to inconsistent visibility in low grayscale regions. Furthermore, OLED panels exhibit spatial dependence in brightness control at low grayscale levels, resulting in significant fluctuations in dark area brightness. If interpolation or dithering algorithms (such as FRC) are not handled properly, brightness flickering or color banding can occur, further reducing detail. In summary, traditional gamma-ray fitting algorithms are insufficient in terms of fitting accuracy, color consistency, and preservation of low grayscale details, requiring more complex piecewise correction or adaptive algorithms for improvement.

[0068] Therefore, there is an urgent need for an automated gamma adjustment method that possesses rapid response, high-precision adjustment capabilities, and adaptability to different display characteristics. This method should improve gamma adjustment efficiency while reducing the degree of human intervention, ensuring the stability and consistency of the adjustment results.

[0069] To address the aforementioned problems and achieve the above objectives, this application provides a grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation. This method can be applied to smart terminals and servers. This application uses the application of this method in a server as an example for illustration, and some other descriptions in the embodiments are illustrative and not intended to limit the scope of protection of this application, and will not be described in detail thereafter. The specific implementation of the method is as follows... Figure 1 As shown:

[0070] Step 101: Convert the acquired color image from RGB space to HSI space, and generate a fuzzy dissimilarity histogram describing the global brightness contrast distribution based on the brightness channel in HSI space.

[0071] Step 102: Calculate the clipping threshold based on the image information of the fuzzy dissimilarity histogram, and reallocate new brightness to the pixels to be adjusted whose brightness is greater than the clipping threshold.

[0072] Step 103: Calculate the gamma value corresponding to the pixel to be adjusted, and use the gamma value to correct the non-linear response of the brightness.

[0073] Step 104: Assign function values ​​to the smooth region and texture detail region of the image based on the nonlinear response, and fuse the brightness of the smooth region and texture detail region using a preset intensity conversion formula and function values.

[0074] Step 105: Calculate the saturation of the fused image, and convert the image including saturation and brightness to RGB space to obtain the final color image.

[0075] The grayscale gamma adjustment method, device, and medium based on fuzzy dissimilarity context intensity transformation provided in this application first constructs a fuzzy dissimilarity histogram based on the input image, which also provides the average difference value for each grayscale level; then, the histogram is truncated to limit over-enhancement; gamma correction is performed to obtain better display fidelity; next, a context intensity transformation is applied to restore the natural features of the image, thereby obtaining the final enhanced image, and the original image and the enhanced image are fused; then, the image saturation is calculated, and the enhanced intensity and saturation are converted back to RGB space. This solves the problems of low efficiency and poor accuracy in grayscale gamma adjustment in the prior art, and achieves fast and accurate grayscale gamma adjustment.

[0076] In one specific embodiment, the specific implementation of generating a fuzzy dissimilarity histogram describing the global luminance contrast distribution based on the luminance channels in the HSI space includes:

[0077] Based on the luminance channel, the following calculation process is performed for each pixel:

[0078] Based on pixel intensity, calculate the similarity value between each pixel and each pixel in the preset neighborhood; calculate the average of the similarity values ​​of all pixels in the preset neighborhood to obtain the similarity index of the pixel; obtain the complement of the similarity index to obtain the dissimilarity index.

[0079] Pixel intensity is a numerical expression of brightness.

[0080] A fuzzy dissimilarity histogram is constructed based on the dissimilarity index.

[0081] The dissimilarity index is used to characterize the difference in intensity between a pixel and its neighboring pixels.

[0082] Specifically, the contrast intensity between a pixel and each pixel in a preset neighborhood is obtained based on the dissimilarity index; a fuzzy dissimilarity histogram is generated based on the cumulative values ​​of different indices of pixels at each brightness level in the image.

[0083] Different brightness levels correspond to different brightness levels, and the same brightness level corresponds to the same brightness level.

[0084] Specifically, firstly, for each pixel in the image, the brightness difference between all pixels in its 3x3 neighborhood and the center pixel is determined, and a similarity value is calculated using a fuzzy membership function (the first membership function). Then, the similarity values ​​of all pixels in the neighborhood are averaged to obtain the similarity index corresponding to that pixel. The complement of this complement is then used to obtain the dissimilarity index, representing the contrast intensity between that pixel and its surrounding environment. Finally, based on the cumulative values ​​of different indices for pixels at each brightness level in the image, a fuzzy dissimilarity histogram describing the global brightness contrast distribution is generated.

[0085] The original color image is denoted as F(x,y)={R(x,y),G(x,y),B(x,y)}, where R, G, and B represent the red, green, and blue channels, respectively, (x,y) are pixel coordinates, x=1,2,...,M, y=1,2,...,N, and M and N are the width and height of the image, respectively. Magnitude stretching is a process performed on each color channel to improve the color saturation of the image. The RGB color model is converted to the HSI model (hue, saturation, brightness) through stretching to observe hue, saturation, and brightness separately. See formula (1) for details:

[0086]

[0087] In the HSI model, the luminance channel I(x,y) is processed separately to increase the contrast of the image.

[0088] A fuzzy dissimilarity histogram (FDH) is constructed based on fuzzy logic and is used for a given input image. Assuming the input image has a size of M×N, that is:

[0089] I={I(x,y)|0≤x≤M-1,0≤y≤N-1}.............(2)

[0090] The intensity (numerical expression of brightness) of each pixel is represented as I(x,y)=g k, where k∈{0,1,...,L-1}, and L represents the total intensity level. The similarity of each pixel in the 3×3 neighborhood is calculated using fuzzy logic. The neighborhood similarity (similarity value) is found by applying fuzzy logic. The similarity value is obtained based on the calculation formula of the membership function (MF), as shown in formula (3):

[0091]

[0092] Where p = x + i, q = y + i, i, j ∈ {-1, 0, 1}, σ represents the standard deviation, and μs(p,q) is the similarity value.

[0093] The membership values ​​of neighboring pixels obtained using fuzzy logic are used to create a fuzzy set called "similarity," which is defined by the fuzzy membership function Ψ. s Definition. The similarity index is calculated using the fuzzy membership function, as shown in formula (4):

[0094]

[0095] The new fuzzy set is called the "different" set, which is obtained by taking the complement of the fuzzy similarity index (similarity index) to obtain the dissimilarity index (contrast). Therefore, the membership function of the new fuzzy set "different" is given by formula (5):

[0096]

[0097] Where μd(x,y) represents the dissimilarity index. Represents Ψ s The supplement to .

[0098] The degree of difference (contrast) of each pixel relative to its neighboring pixels is shown in Equation (5), that is, the numerical expression of the dissimilarity index is used to characterize the contrast intensity.

[0099] Using the new fuzzy set "different", construct a fuzzy dissimilarity histogram H. d See formulas (6), (7) and (8):

[0100] H d ={h d (g k )|0≤g k ≤L-1}..................(6)

[0101]

[0102] Where μ is a value between 0 and 1, and d k Indicates the intensity level.

[0103] Where, dk It is a discrete intensity level in the fuzzy dissimilarity histogram, used to quantify the brightness level of a pixel and to identify areas that need enhancement in the construction of the fuzzy dissimilarity histogram.

[0104] The intensity level and pixel intensity can be in a one-to-one correspondence or a preset range of pixel intensity corresponding to the intensity level. This application does not impose any restrictions, and users can set it according to their actual needs.

[0105] Specifically, contextual data in neighboring pixels of an image is interpreted using a fuzzy dissimilarity histogram (FDH). The FDH provides an average measure of dissimilarity for each intensity level in the input image.

[0106] In one specific embodiment, the image information includes: the total number of pixels, brightness range, and maximum slope of the fuzzy dissimilarity histogram.

[0107] The specific implementation of calculating the shearing threshold based on image information from fuzzy dissimilarity histograms includes:

[0108] Input the total number of pixels, the brightness range, and the maximum slope into the preset shearing determination formula to obtain the shearing threshold output by the shearing determination formula.

[0109] The formula for determining the shearing is shown in formula (9):

[0110]

[0111] Where α represents the clipping threshold; P represents the total number of pixels; Q represents the brightness range; β represents the clipping factor; and S... m This indicates the maximum slope.

[0112] In one specific embodiment, the implementation of reallocating new brightness to pixels whose brightness is greater than the clipping threshold includes:

[0113] Input the pixel intensity of the pixel to be adjusted into a preset mapping formula to obtain a new pixel intensity output by the mapping formula. Then assign the new pixel intensity to the pixel to be adjusted.

[0114] Pixel intensity is a numerical expression of brightness.

[0115] The mapping formula is given in formula (10):

[0116]

[0117] Among them, I c Let l represent the new pixel intensity, l represent the pixel intensity before allocation, and PDF() represent the probability density function. m Indicates the maximum pixel intensity level.

[0118] Specifically, a "shearing" mechanism is designed on the fuzzy dissimilarity histogram to limit excessive contrast contribution. The process involves calculating a shearing threshold based on the total number of pixels, the brightness range, and the maximum slope of the histogram. The portion of the histogram exceeding this threshold is truncated, and the truncated pixels are then evenly distributed back across all brightness levels (equivalent to intensity levels). This prevents excessive contrast concentration at a few brightness levels, maintains the smoothness of the overall brightness mapping, and establishes a new cumulative distribution basis for subsequent brightness remapping.

[0119] The contrast is limited by cropping the peak values ​​in the histogram of the fuzzy dissimilarity. The cropped pixels are then reassigned to new intensity levels. The higher the cropping threshold, the higher the contrast. The cropping threshold is determined using formula (9).

[0120] When β = 0, the clipping threshold is P / Q, and the pixel at that position will remain unchanged. However, as β increases, the contrast will be significantly enhanced. Therefore, the clipping threshold plays an important role in adjusting contrast. This is achieved by using the cumulative distribution function (CDF). We obtain the mapping function (mapping formula) I used to remap image intensity levels. c =T(l)=CDF(l)×l m , where T(l) represents the remapping function.

[0121] In one specific embodiment, the specific implementation of calculating the gamma value corresponding to the pixel to be adjusted includes:

[0122] The pixel intensity of the adjusted pixel is input into the preset gamma parameter calculation formula to obtain the gamma value output by the gamma parameter calculation formula.

[0123] The formula for calculating the gamma parameter is shown in formula (11):

[0124]

[0125] Where γ represents the gamma value and l represents the pixel intensity before allocation.

[0126] In image processing, the gamma factor describes the nonlinearity of intensity.

[0127] In one specific embodiment, the specific implementation of correcting the nonlinear response of brightness using gamma values ​​includes:

[0128] Input the gamma value into the preset nonlinear description formula to obtain the nonlinear response output by the nonlinear description formula.

[0129] The nonlinear description formula is given in formula (12):

[0130]

[0131] Among them, I g (x,y) represents the nonlinear response, i.e., the pixel intensity after gamma correction, located at coordinates (x,y). c This represents the new pixel intensity, where pixel intensity is a numerical expression of brightness, I. cmax γ represents the maximum new pixel intensity, and γ represents the gamma value.

[0132] Specifically, each pixel undergoes initial contrast enhancement, followed by the calculation of an appropriate gamma value using a cumulative distribution function to further correct the non-linear response of brightness. Gamma transform can finely adjust intermediate grayscale levels while preserving details in both highlights and shadows, allowing the image to exhibit richer detail and more natural transitions across different brightness areas.

[0133] This application adjusts the brightness distribution of an image through nonlinear transformation, enhances visual contrast, makes it more in line with the characteristics of human eye perception, and improves user experience.

[0134] In one specific embodiment, the specific implementation of fusing the brightness of smooth areas and texture detail areas using a preset intensity conversion formula and function value includes:

[0135] Input the function value into the intensity conversion formula to obtain the brightness of the blended smooth area and texture detail area output by the intensity conversion formula.

[0136] The intensity conversion formula is given in formula (13):

[0137] I e (x,y)=μc(x,y)I g (x,y)+μc(x,y)I(x,y).............(13)

[0138] Among them, I e (x,y) represents the enhanced strength after fusion, μc(x,y) represents the function value, and I g (x,y) represents the enhanced pixel intensity, and I(x,y) represents the original pixel intensity.

[0139] Specifically, the enhanced image and the original image are fused using formula (13).

[0140] Specifically, to avoid applying the same intensity value uniformly across different scenes, a new membership function (second membership function) representing contextual contrast is constructed based on the similarity index obtained above. This function distinguishes between smooth regions and texture detail regions. Between brightness enhancement and original brightness, the obtained function values ​​(membership values) are weighted and fused. Smooth regions retain more of their original brightness to avoid over-enhancement, while texture detail regions are treated with enhanced brightness. This approach ensures improved contrast while fully preserving the natural features and detail quality of the image.

[0141] Contrast-limited blurring of dissimilarity histograms combined with gamma correction can effectively enhance contrast; however, pixels with the same intensity are converted to the same intensity, which can lead to color distortion. Therefore, a new blur set, called contextual contrast, is constructed, which includes both intensity level and pixel location. This process, of course, preserves the natural characteristics of the input image.

[0142] The second membership function is given by formula (14):

[0143]

[0144] Where μc(x,y) represents the function value. Let x represent the mean and σ represent the standard deviation.

[0145] Contextual contrast assigns higher membership values ​​to smooth regions and vice versa. In the input image, pixels belonging to smooth regions are assigned lower membership values, while texture detail regions are assigned higher membership values. Therefore, contextual information is used for intensity transformation, and the intensity transformation formula is given in formula (13).

[0146] Based on formula (13), the intensity of pixels belonging to smooth regions is close to the original intensity, while the intensity of pixels belonging to texture detail regions is close to the enhanced intensity obtained by the proposed method. Contrast-enhanced images (CIT) do not over-enhance the original or even better contrast images, which is also a desirable characteristic.

[0147] This application integrates local similarity and intensity distribution statistical characteristics through formula (14), dynamically adjusts the function value, provides a data basis for subsequent context intensity transformation, and achieves the purpose of protecting smooth regions, strengthening detailed regions, and balancing global and local aspects.

[0148] In one specific embodiment, the specific implementation of calculating the saturation of the fused image includes:

[0149] Different saturation calculation formulas are used to calculate saturation for pixels of different intensities.

[0150] Specifically, the formula for calculating saturation is shown in formula (15):

[0151]

[0152] Where S(x,y) represents saturation.

[0153] For high-intensity pixels, we have I(x,y)→(L-1), or R,G,B→(L-1), which gives formula (16):

[0154]

[0155] In formula (16), the numerator term approaches 1. Therefore, we obtain formula (17):

[0156]

[0157] For low-intensity pixels, the same procedure is used to reveal the desaturation effect by maximizing the saturation values ​​of both the input and improved images to restore the saturation of the equalized image. Then, the equalized image is merged into the HSI model and converted back to the RGB color model, as shown in formula (18):

[0158]

[0159] Then, the model is converted back to the HSI model to obtain the converted saturation S. e (x,y), maximizing the saturation S extracted from the input image. o (x,y). The final saturation is obtained using formula (19).

[0160] S m (x,y)=max{S o (x,y),S e (x,y)}.................(19)

[0161] The enhanced intensity and maximum saturation are then combined and converted back to the RGB color model. The final output image is shown in Equation (20):

[0162]

[0163] Specifically, this application collects original grayscale images and their corresponding actual brightness responses from various types of displays, then extracts the brightness and chromaticity information of key grayscale points in the images and performs preliminary grayscale correction. A fuzzy algorithm is used to adjust the grayscale gamma for each display type, forming standard grayscale adjustment curves for various types of displays. Specifically, for each display type, grayscale gamma adjustment is performed on the real-time image during actual use (refer to steps 101-105), calculating the difference between the image to be adjusted and the standard image. Based on the known adjustment curve, a fuzzy logic adjustment algorithm is used to adjust the grayscale gamma, quickly approaching the optimal grayscale gamma. The adjustment results are optimized against image contrast, color reproduction, and other indicators. The adjustment results are verified to ensure that the image does not produce color distortion and meets the target display result. If the verification result does not meet the preset standard, the process returns to the grayscale gamma adjustment step. The optimized grayscale parameters are stored and applied to the real-time display system of the display type to achieve automatic grayscale adjustment in batch production. Simultaneously, system stability testing is performed to ensure consistent adjustment accuracy across different display models.

[0164] Ultimately, this enables rapid, accurate, and automated adjustment of grayscale for different displays, significantly improving the visual effect of images, reducing manual intervention, and increasing production efficiency.

[0165] Specifically, through Figure 2 This application will now be described in further detail.

[0166] Specifically, in order to verify the validity of this application, it was verified using real data.

[0167] 1) Summary

[0168] Rapid adjustment of grayscale gamma facilitates quick previewing of adjustments, enabling a faster identification of the most comfortable and natural display effect. However, traditional adjustment processes rely heavily on manual operation, resulting in low efficiency; rapid adjustment of grayscale gamma remains a challenge. This paper addresses this by using empirical parameters and a fuzzy algorithm to quickly approximate the appropriate gamma value.

[0169] 2) Dataset

[0170] This invention selects TID2008 and TID2013 as datasets. The TID2008 dataset contains 25 reference images, 17 distortion types, and 4 distortion intensities. The distortion types include additive Gaussian noise, color component noise, spatially correlated noise, quantization noise, JPEG / JPEG2000 compression and transmission errors, contrast variations, and other categories, totaling 17 types. The TID2013 dataset contains the same 25 reference images as TID2008. Compared to TID2008, it adds 7 distortion types and one additional intensity level. For image quality evaluation metrics, we choose entropy, contrast enhancement factor (CEF), color (C), saturation (S), and Hue deviation index (HDI) to evaluate the quality of existing methods and quantitative methods.

[0171] 3) Results

[0172] As shown in Table 1, the proposed FDGC method achieves a very good trade-off among all image quality evaluation metrics, thus outperforming all other existing contrast enhancement techniques. However, the proposed method has the highest computational complexity compared to all other existing techniques.

[0173]

[0174] Table 1 Performance evaluation of this application and related technologies

[0175] Wherein, Entropy represents entropy, HDI represents hue deviation index, Saturation represents saturation, CEF represents contrast enhancement factor, Gradient represents gradient, Colorfulness represents color, and Time represents time.

[0176] Wherein, Original, FLH, OGCWS, CLAHE-DGC, IAGC, PFT, and AFELCE represent existing baseline methods, and FDGC represents the method provided in this application.

[0177] The entropy value is the highest for the input and all other existing methods. The highest entropy value indicates the richest information present in the image. The hue deviation index (HDI) defines the hue change between the original and enhanced images. A smaller HDI value indicates less hue change. An HDI change of less than 1% indicates that the system preserves hue well, and the enhanced image shows no color distortion. As can be clearly seen from Table 1, the proposed method preserves hue very well compared to other existing methods. Therefore, the proposed method does not exhibit any color distortion. The saturation value of FDGC is almost identical to the original saturation value; therefore, the proposed method is limited to abrupt changes in intensity level.

[0178] Compared to existing methods, the proposed method exhibits higher color value. Therefore, the proposed method is used to reproduce the scene with improved colors. The CEF values ​​obtained through the proposed FDGC method result in significantly greater contrast enhancement values ​​for all five images, as evidenced by the qualitative analysis.

[0179] The uniform and smooth equalization technique results in a higher gradient value. Compared with all other existing methods, the proposed method has a higher gradient value. Therefore, the proposed method can produce smooth images without distortion.

[0180] For details, please refer to Figure 3 The brightness change curve. Figure 3 In the diagram, the horizontal axis represents Input Voltage, and the vertical axis represents Brightness.

[0181] This application first constructs a fuzzy dissimilarity histogram based on the input image, which also provides the average difference value for each gray level. The histogram is then truncated to limit over-enhancement. Gamma correction is performed to obtain better display fidelity. Next, an overlay intensity transform is applied to recover the natural features of the image, thus obtaining the final enhanced image. Finally, experiments were conducted on various color images from different databases, comparing the performance of this application with existing technologies from both subjective and objective perspectives. The final test results show that this application outperforms the output of existing technologies.

[0182] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include a processor 401, a communication interface 402, a memory 403, and a communication bus 404. The processor 401, communication interface 402, and memory 403 communicate with each other via the communication bus 404. The processor 401 can call logical instructions from the memory 403 to execute a grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation.

[0183] Furthermore, the logical instructions in the aforementioned memory 403 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0184] On the other hand, the present invention also provides a computer program product, the computer program product including a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, and when the program instructions are executed by a computer, the computer is able to execute the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation provided by the above methods.

[0185] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation provided in the above embodiments.

[0186] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0187] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0188] Finally, it should be noted that the above descriptions are merely preferred embodiments of this application, and this application is not limited to the above embodiments. It is understood that other improvements and variations directly derived or conceived by those skilled in the art without departing from the spirit and concept of this application should be considered to be included within the protection scope of this application.

Claims

1. A grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation, characterized in that, The method includes: The acquired color image is converted from RGB space to HSI space, and a fuzzy dissimilarity histogram describing the global brightness contrast distribution is generated based on the brightness channel in HSI space. Based on the image information of the fuzzy dissimilarity histogram, a clipping threshold is calculated, and new brightness is reallocated to pixels whose brightness is greater than the clipping threshold. Calculate the gamma value corresponding to the pixel to be adjusted, and use the gamma value to correct the nonlinear response of the brightness; Based on the nonlinear response, function values ​​are assigned to the smooth regions and texture detail regions of the image, and the brightness of the smooth regions and texture detail regions is fused using a preset intensity conversion formula and the function values. The saturation of the fused image is calculated, and the image, which includes both saturation and brightness, is converted to the RGB space to obtain the final color image.

2. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, A fuzzy dissimilarity histogram describing the global luminance contrast distribution is generated based on the luminance channels in the HSI space, including: Based on the luminance channel, the following calculation process is performed for each pixel: Based on pixel intensity, calculate the similarity value between the pixel and each pixel in the preset neighborhood; calculate the average of the similarity values ​​of all pixels in the preset neighborhood to obtain the similarity index of the pixel; obtain the complement of the similarity index to obtain the dissimilarity index, where pixel intensity is a numerical expression of brightness; A fuzzy dissimilarity histogram is constructed based on the dissimilarity index, where the dissimilarity index is used to characterize the difference in intensity between a pixel and its neighboring pixels.

3. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, Image information includes: the total number of pixels, brightness range, and maximum slope of the fuzzy dissimilarity histogram; Calculating the clipping threshold based on the image information of the fuzzy dissimilarity histogram includes: Input the total number of pixels, the brightness range, and the maximum slope into a preset shearing determination formula to obtain the shearing threshold output by the shearing determination formula; The shear determination formula includes: Where α represents the clipping threshold; M represents the total number of pixels; N represents the brightness range; β represents the clipping factor; and S... m This indicates the maximum slope.

4. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, Reassigning new brightness to pixels whose brightness is greater than the clipping threshold includes: The pixel intensity of the pixel to be adjusted is input into a preset mapping formula to obtain a new pixel intensity output by the mapping formula, where the pixel intensity is a numerical expression of brightness; the new pixel intensity is then assigned to the pixel to be adjusted. The mapping formulas include: Among them, I c Let l represent the new pixel intensity, l represent the pixel intensity before assignment, and PDF(.) represent the probability density function. m Indicates the maximum pixel intensity level.

5. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, Calculating the gamma value corresponding to the pixel to be adjusted includes: The pixel intensity of the adjusted pixel is input into a preset gamma parameter calculation formula to obtain the gamma value output by the gamma parameter calculation formula. The formula for calculating gamma parameters includes: Where γ represents the gamma value and l represents the pixel intensity before allocation.

6. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, Correcting the nonlinear response of brightness using the gamma value includes: The gamma value is input into a preset nonlinear description formula to obtain the nonlinear response output by the nonlinear description formula. The nonlinear description formula includes: Among them, I g (x,y) represents the nonlinear response, I c This represents the new pixel intensity, where pixel intensity is a numerical expression of brightness, I. cmax γ represents the maximum new pixel intensity, and γ represents the gamma value.

7. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, The brightness of the smooth area and the texture detail area are fused using a preset intensity conversion formula and the function value, including: The function value is input into the intensity conversion formula to obtain the brightness of the blended smooth area and texture detail area output by the intensity conversion formula; The intensity conversion formula includes: I e (x,y)=μc(x,y)I g (x,y)+μc(x,y)I(x,y); Among them, I e (x,y) represents the enhanced strength after fusion, μc(x,y) represents the function value, and I g (x,y) represents the enhanced pixel intensity, and I(x,y) represents the original pixel intensity.

8. The grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation according to claim 1, characterized in that, Calculate the saturation of the fused image, including: Different saturation calculation formulas are used to calculate saturation for pixels of different intensities.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation as described in any one of claims 1 to 8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the grayscale gamma adjustment method based on fuzzy dissimilarity context intensity transformation as described in any one of claims 1 to 8.

Images