A high-precision calibration method in TFT display production

Abstract

The present invention provides a high-precision calibration method in the production of TFT displays, which relates to the field of TFT display calibration, including: establishing particle groups of various types of TFT displays respectively, obtaining the optimal intermediate values ​​of brightness and contrast of various types of TFT displays, i.e., initial values, by simulating the movement of particles in the search space, establishing gene libraries of various types of TFT displays respectively, adjusting the brightness and contrast of various types of TFT displays by selection, crossover and mutation, and iteratively screening to obtain the brightness and contrast of the display reaching a predetermined standard value, using a deep learning color correction model to adjust the color of various types of TFT displays, and using a Markov random field algorithm to adjust the brightness and color uniformity of various areas of the TFT display screen according to the calibrated color parameters. The present invention aims to achieve accurate personalized calibration for different types of TFT displays.

Description

Technical Field

[0001] The invention relates to the field of TFT display calibration, and in particular to a high-precision calibration method in TFT display production. Background Art

[0002] TFT display, especially TFT-LCD (Thin Film Transistor Liquid Crystal Display), is one of the most widely used display technologies. Its display principle is based on the photoelectric effect of liquid crystal molecules. It changes the polarization state of light by controlling the arrangement direction of liquid crystal molecules, thereby realizing the display of images. Each pixel in the TFT display is controlled by a thin film transistor, which enables each pixel to be switched on and off and grayscale adjusted independently, thereby improving the contrast and color richness of the display.

[0003] In the production process of TFT (Thin Film Transistor) displays, calibration is a crucial link, which is directly related to the display effect and user experience of the display;

[0004] However, different models of TFT displays have differences in display performance, and existing methods often use a unified calibration standard, which makes it impossible to perform personalized calibration for different models. Summary of the invention

[0005] The purpose of the embodiment of the present invention is to provide a high-precision calibration method in the production of TFT displays, aiming to achieve accurate and personalized calibration for different models of TFT displays.

[0006] In one aspect, the present invention provides a high-precision calibration method in TFT display production, comprising:

[0007] Establish particle swarms for each type of TFT display respectively, and obtain the optimal intermediate value of brightness and contrast of each type of TFT display by simulating the movement of particles in the search space, that is, the initial value, wherein each particle in the particle swarm represents the brightness and contrast of a group of TFT displays; determine the particles representing the brightness and contrast of each group of TFT displays, calculate the fitness of the brightness and contrast of each particle, determine the optimal fitness of each particle and the global optimal fitness according to the fitness of each pair of particles, and update the speed and position of each particle to obtain the optimal intermediate value of brightness and contrast of each type of TFT display, that is, the initial value;

[0008] According to the initial values, gene libraries of various models of TFT displays are respectively established, and the brightness and contrast of various models of TFT displays are adjusted and iteratively screened through selection, crossover and mutation to obtain the brightness and contrast of various models of TFT displays that reach the predetermined standard values, that is, the calibrated brightness and contrast of various models of TFT displays, wherein the gene codes in the gene library represent different brightness and contrast of various models of TFT displays;

[0009] According to the calibrated brightness and contrast of each model of TFT display, the color of the TFT display is adjusted using a deep learning color correction model to obtain color parameters that reach a predetermined standard value, i.e., the calibrated color parameters;

[0010] According to the calibrated color parameters, the Gamma value is adjusted according to the brightness and color distribution of the displayed content using an adaptive neural network model to obtain a calibrated Gamma value;

[0011] According to the calibrated Gamma value, the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen of each model to obtain uniform calibrated brightness and color; each area of ​​the TFT display screen is regarded as each node, the potential energy function formula of each node is determined, and the objective function of the Markov random field algorithm is constructed according to the potential energy function of each node. The objective function of the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen to obtain uniform calibrated brightness and color;

[0012] According to the uniformly calibrated brightness and color, the image analysis algorithm is used to analyze the display effects of various types of TFT displays to obtain analysis results;

[0013] According to the analysis result, a comparison is performed with a predetermined standard value to obtain a comparison result. If the comparison result exceeds or reaches a threshold, recalibration is performed. If the comparison result does not exceed the threshold, the calibration is completed.

[0014] Furthermore, a particle swarm of each type of TFT display is established, and the optimal intermediate value of the brightness and contrast of each type of TFT display, i.e., the initial value, is obtained by simulating the movement of particles in the search space, wherein each particle in the particle swarm represents the brightness and contrast of a group of TFT displays, including:

[0015] Determine particles representing the brightness and contrast of each group of TFT displays, each particle representing the brightness and contrast of a group of TFT displays;

[0016] use Calculate the fitness of each particle's brightness and contrast, where is the target brightness, is the target contrast, is brightness, is the contrast;

[0017] According to the fitness of each pair of particles, the optimal fitness of each particle and the global optimal fitness are determined;

[0018] According to the optimal fitness of each particle and the global optimal fitness, use , The speed and position of each particle are updated to obtain the optimal intermediate value of brightness and contrast of each type of TFT display, that is, the initial value, where: and are the velocity and position of particle i, is the inertia weight, and are learning factors, which control the speed at which particles learn to the optimal solution and the global optimal solution respectively. is a random number between [0,1], is the optimal position of particle i, and gbest is the global optimal position.

[0019] Further, according to the initial values, gene libraries of various types of TFT displays are established respectively, and the brightness and contrast of various types of TFT displays are adjusted and iteratively screened through selection, crossover and mutation to obtain the brightness and contrast of the display reaching the predetermined standard value, that is, the calibrated brightness and contrast of various types of TFT displays, wherein the gene codes in the gene library represent different brightness and contrast of various types of TFT displays, including:

[0020] According to the initial values, the gene library of each type of TFT display is established respectively, and the gene codes in the gene library represent different brightness and contrast;

[0021] According to the fitness of each pair of particles, determine the parent particle;

[0022] According to the particles of the parent generation, two parent generation particles are randomly selected to generate two child generation particles in a double-point crossover manner to obtain a number of child generation particles;

[0023] According to the offspring particles, the gene positions are randomly changed to obtain the mutated particles;

[0024] According to the mutated particles, fitness evaluation is performed, and particles whose fitness reaches or exceeds the threshold are selected to enter the next generation;

[0025] The above steps are iterated until the brightness and contrast of each model of TFT display reaching the predetermined standard value, ie, the calibrated brightness and contrast, are obtained.

[0026] Furthermore, according to the calibrated brightness and contrast of each model of TFT display, the color of the TFT display is adjusted using a deep learning color correction model to obtain color parameters that reach a predetermined standard value, that is, calibrated color parameters, including:

[0027] Acquiring color parameters of the TFT display and color parameters of predetermined standard values;

[0028] Build a deep learning model using The difference between the color parameters of the TFT display and the color parameters of the predetermined standard values ​​is calculated, wherein: z is the loss value, is the sample size, It is The target color parameters for samples, is the color parameter predicted by the model;

[0029] According to the difference between the color parameters of the TFT display and the color parameters of the predetermined standard values, the deep learning model is optimized using a back propagation algorithm to obtain an optimized deep learning model;

[0030] According to the color parameters of the TFT display and the color parameters of the predetermined standard values, the color parameters of the TFT display are calibrated using a deep learning model to obtain color parameters that reach the predetermined standard values, that is, calibrated color parameters.

[0031] Further, according to the calibrated color parameters, the Gamma value is adjusted according to the brightness and color distribution of the displayed content using an adaptive neural network model to obtain a calibrated Gamma value, including:

[0032] Obtaining the Gamma value of the TFT display and the Gamma value of a predetermined standard value;

[0033] According to the Gamma value of the TFT display and the Gamma value of the predetermined standard value, use , Gamma value adjustment and calibration are performed to obtain a calibrated Gamma value, where: is the first input feature Quantity, It is Input the components to The weight of a neuron, is the bias term, It is The net input to each neuron is is the activation function, is the output of the f-th neuron.

[0034] Furthermore, according to the calibrated Gamma value, the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​each model of TFT display screen to obtain uniformly calibrated brightness and color, including:

[0035] Obtaining brightness and color characteristics of each area of ​​the TFT display screen, wherein the characteristics include average brightness, brightness standard deviation, color mean, and color variance;

[0036] Treat each area of ​​the TFT display screen as each node and determine the neighborhood relationship between the nodes;

[0037] According to the neighborhood relationship between nodes, the potential energy function formula of each node is determined, that is, To obtain the potential energy function of each node, the potential energy function includes a data term and a smoothing term, where: is the total potential energy function, is the data item potential energy function of node g, is the smooth potential function between nodes g and h, N is the set of neighborhood relations, is the state of node g, where g and h are two adjacent nodes respectively;

[0038] According to the potential energy function of each node, the objective function of the Markov random field algorithm is constructed, that is, ,in, is the objective function, Represents the search for the node state configuration that minimizes the total potential energy ;

[0039] The objective function of the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen to obtain uniformly calibrated brightness and color.

[0040] Further, according to the uniformly calibrated brightness and color, an image analysis algorithm is used to analyze the display effect of the TFT display to obtain analysis results, including:

[0041] Based on the uniformly calibrated brightness and color image, use Calculate the contrast of the image, where is the contrast, is the brightness change, is brightness;

[0042] Based on the uniformly calibrated brightness and color image, use Calculate the image clarity index, where is the image function, is the clarity index, Indicates the change in the curvature of the image in the horizontal direction. It reflects the change of curvature of the image in the vertical direction;

[0043] Based on the uniformly calibrated brightness and color image, use Calculate the color difference between different regions or pixels in the image to obtain the color consistency parameter, where: is the color consistency parameter, and are the red, green, and blue component values ​​of the two pixels respectively;

[0044] According to the image contrast, clarity index, and color consistency parameters, statistics are calculated to obtain analysis results.

[0045] Through the above technical scheme, the high-precision calibration method in the production of TFT displays provided by the present invention realizes accurate personalized calibration of different models of TFT displays; the particle swarm algorithm is used to find the optimal initial values ​​of brightness and contrast for each model of TFT displays, which lays a solid foundation for subsequent calibration work, and iterative screening is combined with the gene library algorithm to further improve the accuracy of calibration, ensuring that the brightness and contrast of the display can reach the predetermined standard value; the present invention establishes a particle swarm and a gene library for each model of TFT display, fully considers the differences in display performance of different models of displays, realizes personalized calibration, and improves the pertinence and effectiveness of calibration; the deep learning color correction model is used to calibrate the display The color of the display can be adjusted to restore the color more accurately, improve the color saturation, contrast and color accuracy, and make the color performance of the display more outstanding; the Gamma value is dynamically adjusted according to the brightness and color distribution of the display content through the adaptive neural network model, ensuring that the display can maintain the best display effect under different display contents, and improving the adaptability and flexibility of the display; the Markov random field algorithm is used to finely adjust the brightness and color uniformity of each area of ​​the display screen, effectively solving the problem of inconsistent brightness and color of various parts of the screen, and improving the consistency of the overall display effect; the present invention analyzes the calibrated display effect through an image analysis algorithm and compares it with a predetermined standard value. If the comparison result does not meet the requirements, recalibration is performed to form a closed-loop calibration process to ensure the accuracy and reliability of the calibration result.

[0046] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0048] Figure 1 It is a schematic diagram of a high-precision calibration method in TFT display production provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0050] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application are in compliance with the relevant provisions of national laws and regulations. In the embodiments of this application, some existing solutions in the industry such as certain software, components, and models may be mentioned, which should be considered as exemplary. Their purpose is only to illustrate the feasibility of implementing the technical solution of this application, but it does not mean that the applicant has or will necessarily use the solution.

[0051] like Figure 1 As shown, the high-precision calibration method in TFT display production provided by the embodiment of the present invention includes:

[0052] 11. Establish particle swarms of various types of TFT displays respectively, and obtain the optimal intermediate values ​​of brightness and contrast of various types of TFT displays, i.e., initial values, by simulating the movement of particles in the search space, wherein each particle in the particle swarm represents the brightness and contrast of a group of TFT displays;

[0053] 12. According to the initial values, gene libraries of various types of TFT displays are established respectively, and the brightness and contrast of various types of TFT displays are adjusted and iteratively screened through selection, crossover and mutation to obtain the brightness and contrast of the display reaching the predetermined standard value, that is, the calibrated brightness and contrast, wherein the gene codes in the gene library represent different brightness and contrast;

[0054] 13. According to the calibrated brightness and contrast, the deep learning color correction model is used to adjust the color of each model of TFT display to obtain color parameters that reach the predetermined standard value, that is, the calibrated color parameters;

[0055] 14. According to the calibrated color parameters, the Gamma value is adjusted according to the brightness and color distribution of the displayed content using an adaptive neural network model to obtain a calibrated Gamma value;

[0056] 15. According to the calibrated Gamma value, the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen to obtain uniformly calibrated brightness and color;

[0057] 16. According to the uniformly calibrated brightness and color, use the image analysis algorithm to analyze the display effects of various types of TFT displays to obtain the analysis results;

[0058] 17. According to the analysis result, the comparison is performed with the predetermined standard value to obtain the comparison result. If the comparison result exceeds or reaches the threshold, recalibration is performed. If the comparison result does not exceed the threshold, the calibration is completed.

[0059] In an embodiment of the present invention, a particle swarm is established to find the optimal initial values ​​of brightness and contrast. By simulating the movement of particles in the search space, the optimal intermediate values ​​(i.e., initial values) of the brightness and contrast of each type of TFT display are found, which can efficiently explore the possible solution space and quickly converge to a better initial parameter setting; a good starting point is provided for the subsequent establishment of a gene library and optimization of a genetic algorithm, which helps to speed up the speed and accuracy of the entire calibration process; a gene library is established and the brightness and contrast are adjusted by a genetic algorithm, and the brightness and contrast of the display are iteratively screened and optimized by using the selection, crossover, and mutation operations in the genetic algorithm to gradually approach the predetermined standard value. Through the evolution process of the genetic algorithm, the brightness and contrast parameters of the display can be gradually optimized to make it more in line with the preset calibration standard; a deep learning color correction model is used to adjust the color. After determining the calibrated brightness and contrast, the deep learning model is used to further adjust the color of the display to make it reach the predetermined standard value. The deep learning model can capture complex color mapping relationships, thereby achieving high-precision color correction and improving the color performance of the display; an adaptive neural network model is used to adjust the Gamma value according to the display content. The brightness and color distribution of the display are adjusted dynamically using the adaptive neural network model to optimize the display effect of the display; by adaptively adjusting the Gamma value, it can be ensured that the display can present the best visual effect under different display contents; the Markov random field algorithm is used to adjust the brightness and color uniformity, and the Markov random field algorithm is used to perform global optimization adjustment for the brightness and color non-uniformity problems in each area of ​​the TFT display screen; through the modeling and optimization of the Markov random field algorithm, the brightness and color uniformity of the display screen can be significantly improved, and the overall display quality can be improved; the image analysis algorithm is used to evaluate the calibration effect. After completing all the calibration steps, the image analysis algorithm is used to comprehensively evaluate the display effect of the display. The image analysis algorithm can objectively measure the display effect of the calibrated display, providing a reliable basis for subsequent comparison and recalibration; the analysis results are compared with the predetermined standard values ​​to decide whether to recalibrate. The analysis results are compared with the predetermined standard values ​​to determine whether the calibration effect meets the requirements. The comparison results can timely discover the deficiencies and deviations in the calibration process, and decide whether to recalibrate as needed to ensure that the final calibration effect meets the preset standards.

[0060] like Figure 1 As shown, 11. Establish particle swarms of various types of TFT displays respectively, and obtain the optimal intermediate values ​​of brightness and contrast of various types of TFT displays, i.e., initial values, by simulating the movement of particles in the search space, wherein each particle in the particle swarm represents the brightness and contrast of a group of TFT displays, including:

[0061] Determine particles representing the brightness and contrast of each group of TFT displays, each particle representing the brightness and contrast of a group of TFT displays;

[0062] use Calculate the fitness of each particle's brightness and contrast, where is the target brightness, is the target contrast, is brightness, is the contrast;

[0063] According to the fitness of each pair of particles, the optimal fitness of each particle and the global optimal fitness are determined;

[0064] According to the optimal fitness of each particle and the global optimal fitness, use , The speed and position of each particle are updated to obtain the optimal intermediate value of brightness and contrast of each type of TFT display, that is, the initial value, where: and are the velocity and position of particle i, is the inertia weight, and are learning factors, which control the speed at which particles learn to the optimal solution and the global optimal solution respectively. is a random number between [0,1], is the optimal position of particle i, and gbest is the global optimal position.

[0065] In the embodiment of the present invention, the particles representing the brightness and contrast of each pair of displays are determined. The particle swarm optimization algorithm first needs to represent each possible solution in the problem space as a "particle". In the calibration problem, each particle represents a pair of specific display brightness and contrast values. By defining particles, the algorithm can transform the complex search problem into a problem of finding the optimal solution in the particle space. The fitness function is used to calculate the fitness of each pair of particles. The fitness function is a standard for evaluating the quality of particles. In this calibration problem, the fitness function may be based on the evaluation indicators of the display effect, such as clarity, contrast, color reproduction, etc. By calculating each pair of particles, the particle swarm optimization algorithm can be used to calculate the optimal solution of the particle space. The algorithm can understand the fitness of each particle and the extent to which the brightness and contrast values ​​represented by each particle are close to the optimal solution. According to the fitness of each pair of particles, the individual optimal fitness and the global optimal fitness are determined. The individual optimal fitness refers to the best fitness value encountered by each particle during the iteration process, representing the optimal solution of the particle itself. The global optimal fitness refers to the best fitness value encountered by the entire particle group during the iteration process, representing the optimal solution of the entire search space. By determining the individual optimal and global optimal, the algorithm can guide the particles to move in a more optimal direction. According to the individual optimal fitness and the global optimal fitness, the formula is used to calculate the speed and Position update, the core of the particle swarm optimization algorithm lies in the speed and position update formula of the particle, which updates the speed and position of each particle based on the information of individual optimal and global optimal, as well as a certain degree of randomness; by continuously updating the position and speed of particles, the algorithm can simulate the movement of particles in the search space and gradually approach the optimal solution; efficient search, the particle swarm optimization algorithm can search a larger solution space in a shorter time by simulating the group behavior of particles, thus improving the search efficiency; global optimization, since the algorithm considers the information of individual optimal and global optimal at the same time, it can effectively avoid falling into the local optimal solution and achieve global optimization; strong robustness, The particle swarm optimization algorithm is insensitive to initial conditions and can converge to similar optimal solutions even under different initial conditions, showing strong robustness. It is easy to implement. The principle of the particle swarm optimization algorithm is simple and easy to understand, and it is relatively easy to implement, and is applicable to various complex optimization problems. In the high-precision calibration method in TFT display production, the particle swarm optimization algorithm is used to automatically search and determine the optimal intermediate values ​​of the brightness and contrast of the display, which not only improves the efficiency and accuracy of the calibration, but also provides a good starting point for subsequent calibration steps. The algorithm can quickly locate the brightness and contrast values ​​close to the optimal value, laying a solid foundation for subsequent fine-tuning.

[0066] like Figure 1As shown, 12. According to the initial values, gene libraries of various types of TFT displays are established respectively, and the brightness and contrast of various types of TFT displays are adjusted and iteratively screened through selection, crossover and mutation to obtain the brightness and contrast of the display reaching the predetermined standard value, that is, the calibrated brightness and contrast, wherein the gene codes in the gene library represent different brightness and contrast, including:

[0067] According to the initial values, the gene library of each type of TFT display is established respectively, and the gene codes in the gene library represent different brightness and contrast;

[0068] According to the fitness of each pair of particles, determine the parent particle;

[0069] According to the particles of the parent generation, two parent generation particles are randomly selected to generate two child generation particles in a double-point crossover manner to obtain a number of child generation particles;

[0070] According to the offspring particles, the gene positions are randomly changed to obtain the mutated particles;

[0071] According to the mutated particles, fitness evaluation is performed, and particles whose fitness reaches or exceeds the threshold are selected to enter the next generation;

[0072] The above steps are iterated until the brightness and contrast of the display reach predetermined standard values, that is, the calibrated brightness and contrast are obtained.

[0073] In the embodiment of the present invention, according to the initial value, the gene library of each model of TFT display is established respectively, and the gene code in the gene library represents different brightness and contrast; the particle as the parent generation is determined, based on the initial value obtained by the particle swarm optimization algorithm, that is, the optimal intermediate value of brightness and contrast, the genetic algorithm first selects the particle with excellent performance as the parent generation according to the fitness of each pair of particles, and the parent generation particles carry high-quality gene information, which will serve as the basis for subsequent evolution; the daughter generation particles are generated by randomly selecting two parent generation particles, and according to the double-point crossover method, that is, selecting two intersection points in the gene sequences of the two parent generation particles, and exchanging the gene fragments between the two points, so as to generate two daughter generation particles; the process It simulates the hybridization phenomenon in biological evolution, which helps to produce offspring with new characteristics; gene mutation, which randomly changes the gene positions of the generated offspring particles, is called mutation operation; mutation is an important mechanism in biological evolution, which increases the diversity of the population, helps to jump out of the local optimal solution and explore a broader solution space; fitness evaluation and selection, which evaluates the fitness of the mutated particles and selects particles whose fitness reaches or exceeds a certain threshold to enter the next generation; the process simulates the "survival of the fittest" principle in natural selection, ensuring that only particles with excellent performance can continue to participate in evolution; iterative evolution, repeating the above steps, i.e. selecting parents, generating offspring, gene mutation, fitness evaluation and selection, until The brightness and contrast of the display that meet the predetermined standard values ​​are obtained; the process simulates the iterative process of biological evolution, and gradually approaches the optimal solution through continuous evolution and screening; the global optimization ability is strong, and the genetic algorithm can search for the global optimal solution or the approximate optimal solution in a larger solution space by simulating the biological evolution process; it is particularly important for the calibration of the brightness and contrast of TFT displays, because such problems usually have complex nonlinear relationships; the search efficiency is high, and the genetic algorithm can quickly generate and evaluate a large number of candidate solutions through operations such as selection, crossover and mutation, thereby improving the search efficiency; it is of great significance for the rapid calibration needs in the production environment; it is strong in robustness, and the genetic algorithm has strong initial It is insensitive to initial conditions and parameter settings, and can converge to similar optimal solutions even under different initial conditions or parameter settings; the algorithm has strong robustness and stability in practical applications; it is easy to combine with other algorithms. As a heuristic search algorithm, genetic algorithm is easy to combine with other optimization algorithms to form a hybrid algorithm to further improve the accuracy and efficiency of calibration; in summary, the genetic algorithm is used to perform high-precision calibration of the brightness and contrast of various models of TFT displays, which not only improves the accuracy and efficiency of the calibration of various models of TFT displays, but also enhances the robustness and stability of the algorithm; it provides an effective solution to the calibration problem of various models of TFT displays in the production environment.

[0074] like Figure 1As shown, 13. According to the calibrated brightness and contrast, the color of each model of TFT display is adjusted using a deep learning color correction model to obtain color parameters that reach a predetermined standard value, that is, the calibrated color parameters, including:

[0075] Acquiring color parameters of the TFT display and color parameters of predetermined standard values;

[0076] Build a deep learning model using The difference between the color parameters of the TFT display and the color parameters of the predetermined standard values ​​is calculated, wherein: z is the loss value, is the sample size, It is The target color parameters for samples, is the color parameter predicted by the model;

[0077] According to the difference between the color parameters of the TFT display and the color parameters of the predetermined standard values, the deep learning model is optimized using a back propagation algorithm to obtain an optimized deep learning model;

[0078] According to the color parameters of the TFT display and the color parameters of the predetermined standard values, the color parameters of the TFT display are calibrated using a deep learning model to obtain color parameters that reach the predetermined standard values, that is, calibrated color parameters.

[0079] In an embodiment of the present invention, current color parameters of the TFT display and color parameters of predetermined standard values ​​are obtained; the parameters generally include key indicators such as color gamut, chromaticity, brightness, contrast, etc., which together determine the color expression capability of the display; a deep learning model is constructed, and then a deep learning model is constructed that can learn and understand the complex relationship between color parameters; by using a loss function formula, the model can quantify the difference between the color parameters of the TFT display and the predetermined standard values, providing clear guidance for the subsequent optimization process; using a back propagation algorithm, the deep learning model is iteratively optimized according to the calculated color parameter difference; during the process, the model will continuously adjust its internal parameters to minimize the loss function, thereby gradually approaching the color parameters of the predetermined standard values; color calibration, finally, the color parameters of the TFT display are calibrated using the optimized deep learning model. The model will output calibrated color parameters that reach the predetermined standard values ​​based on the input color parameters, namely the calibrated color parameters; improve color accuracy. Through the fine adjustment of the deep learning color correction model, the color parameters of the display can more accurately match the predetermined standard values, thereby significantly improving the color accuracy; enhance color consistency. The calibrated color parameters can ensure that the color performance of the display remains consistent in different environments and at different times, avoiding visual fatigue or misunderstanding caused by color deviation; improve user experience. Accurate color performance can bring users a more realistic and vivid visual experience; whether watching videos, browsing pictures or doing graphic design, users can feel more delicate and rich color levels; improve production efficiency. The automated and intelligent color calibration process can greatly shorten the calibration time and improve production efficiency; at the same time, it reduces the errors caused by human intervention and ensures the stability and reliability of the calibration results; enhance product competitiveness. High-precision color calibration makes the TFT display more outstanding in color performance, thereby enhancing the market competitiveness of the product; In summary, using the deep learning color correction model to perform high-precision calibration of the color of the TFT display can not only significantly improve the color accuracy and consistency, but also bring users a more outstanding visual experience, while improving production efficiency and product competitiveness.

[0080] like Figure 1 As shown, 14. According to the calibrated color parameters, the Gamma value is adjusted according to the brightness and color distribution of the displayed content using an adaptive neural network model to obtain a calibrated Gamma value, including:

[0081] Obtaining the Gamma value of the TFT display and the Gamma value of a predetermined standard value;

[0082] According to the Gamma value of the TFT display and the Gamma value of the predetermined standard value, use , Gamma value adjustment and calibration are performed to obtain a calibrated Gamma value, where: is the first input feature Quantity, It is Input the components to The weight of a neuron, is the bias term, It is The net input to each neuron is is the activation function, is the output of the f-th neuron.

[0083] In an embodiment of the present invention, the Gamma value is dynamically adjusted. The adaptive neural network model can adjust the Gamma value in real time and dynamically according to the brightness and color distribution of the displayed content; this means that the display can automatically adjust its color performance according to different image contents to adapt to different visual needs; improve color accuracy. By accurately adjusting the Gamma value, the adaptive neural network model can ensure that the display can present accurate colors under various display conditions; it helps to reduce color distortion and deviation, and improve color restoration and saturation; enhance contrast. The adjustment of the Gamma value can also affect the contrast performance of the display; by optimizing the Gamma value, the adaptive neural network model can enhance the contrast of the image, making the dark details and the bright details richer and clearer; optimize the visual experience. The adaptive neural network model adjusts the Gamma value by comprehensively considering the brightness and color distribution of the displayed content, thereby optimizing the user's visual experience; it helps to reduce visual fatigue and misunderstanding, and improve the user's understanding and appreciation of the displayed content; improve display quality. The Gamma value adjusted by the adaptive neural network model can significantly improve the display Improve the display quality of the display; enhance adaptability. The adaptive neural network model can dynamically adjust the Gamma value according to the real-time changes of the display content, which enhances the adaptability of the display to different display contents; in bright scenes or dim scenes, the display can present the best color effect; improve user satisfaction. Accurate color performance and optimized contrast can significantly improve the user's visual experience, thereby enhancing user satisfaction and loyalty to the display; it is an important competitive advantage for display manufacturers; reduce calibration costs. By using the adaptive neural network model to adjust the Gamma value, manufacturers can reduce their dependence on professional calibration equipment and personnel, thereby reducing calibration costs and time; at the same time, because the model can adjust the Gamma value in real time and dynamically, it also reduces the need for frequent calibration caused by changes in display content; in summary, using the adaptive neural network model to adjust the Gamma value according to the brightness and color distribution of the display content can significantly improve the display quality of the TFT display, enhance adaptability, improve user satisfaction, and reduce calibration costs; it plays a vital role in high-precision calibration methods.

[0084] like Figure 1 As shown in FIG15, according to the calibrated Gamma value, the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​each model of TFT display screen to obtain uniformly calibrated brightness and color, including:

[0085] Obtaining brightness and color characteristics of each area of ​​the TFT display screen, wherein the characteristics include average brightness, brightness standard deviation, color mean, and color variance;

[0086] Treat each area of ​​the TFT display screen as each node and determine the neighborhood relationship between the nodes;

[0087] According to the neighborhood relationship between nodes, the potential energy function formula of each node is determined, that is, To obtain the potential energy function of each node, the potential energy function includes a data term and a smoothing term, where: is the total potential energy function, Is a node The data term potential energy function, Is a node and h , is the state of node g, where g and h are two adjacent nodes respectively;

[0088] According to the potential energy function of each node, the objective function of the Markov random field algorithm is constructed, that is, ,in, is the objective function, Represents the search for the node state configuration that minimizes the total potential energy ;

[0089] The objective function of the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen to obtain uniformly calibrated brightness and color.

[0090] In the embodiment of the present invention, the screen features are accurately captured. By acquiring the brightness and color features of each area of ​​the TFT display screen, the algorithm can comprehensively and accurately grasp the overall performance of the screen and the differences that exist; the node and neighborhood relationship is constructed, each area of ​​the screen is regarded as an independent node, and the neighborhood relationship between them is clarified, which lays a solid foundation for the construction of the Markov random field; the setting of the node and neighborhood relationship helps the algorithm understand the mutual influence between the various areas of the screen; the potential energy function is defined, and the potential energy function is carefully set for each node based on the neighborhood relationship between the nodes. The function comprehensively considers the data term and the smoothing term, that is, the actual brightness and color features of each area of ​​the screen and the continuity of the brightness and color between adjacent areas, which provides clear guidance for subsequent adjustments; the objective function is constructed, and the objective function of the Markov random field algorithm is constructed based on the potential energy function of each node; the function comprehensively reflects the overall brightness of the screen The optimization goals of color uniformity are important bases for iterative optimization of the algorithm; uniformity adjustment, using the constructed objective function, the Markov random field algorithm can efficiently adjust the brightness and color of each area of ​​the screen to make it uniform; the adjustment process not only improves the overall display effect of the display, but also ensures the visual comfort of the user when watching; significantly improves uniformity, after fine-tuning the Markov random field algorithm, the brightness and color uniformity of the TFT display screen are significantly improved; both the center and edge areas of the screen can present a consistent and high-quality display effect; optimize the visual experience, uniform brightness and color performance can greatly reduce the user's visual fatigue and discomfort, and provide users with a more natural and comfortable viewing experience; it is especially important for users who use the display for a long time; enhance color accuracy, while ensuring uniformity, the algorithm also pays attention to color accuracy. Through fine adjustment, the colors of each area of ​​the screen are closer to the real value, and the color reproduction ability of the display is improved; the product competitiveness is improved. High-quality display effects and user experience are important competitive advantages of display products. The optimization of brightness and color uniformity by the Markov random field algorithm significantly improves the product quality and market competitiveness of TFT displays; the subsequent maintenance cost is reduced. Uniform brightness and color performance help reduce the failure rate and maintenance requirements of the display, thereby reducing the user's subsequent maintenance cost; at the same time, it also improves user satisfaction and loyalty to the product; in summary, the use of Markov random field algorithm to adjust the brightness and color uniformity of each area of ​​the TFT display screen not only significantly improves the display effect and user experience of the display, but also enhances the competitiveness of the product and reduces the subsequent maintenance cost; it plays a vital role in high-precision calibration methods.

[0091] like Figure 1 As shown in 16, according to the uniformly calibrated brightness and color, the image analysis algorithm is used to analyze the display effect of the TFT display to obtain analysis results, including:

[0092] Based on the uniformly calibrated brightness and color image, use Calculate the contrast of the image, where is the contrast, is the brightness change, is brightness;

[0093] Based on the uniformly calibrated brightness and color image, use Calculate the image clarity index, where is the image function, is the clarity index, Indicates the change in the curvature of the image in the horizontal direction. It reflects the change of curvature of the image in the vertical direction;

[0094] Based on the uniformly calibrated brightness and color image, use Calculate the color difference between different regions or pixels in the image to obtain the color consistency parameter, where: is the color consistency parameter, and are the red, green, and blue component values ​​of the two pixels respectively;

[0095] According to the image contrast, clarity index, and color consistency parameters, statistics are calculated to obtain analysis results.

[0096] In the embodiment of the present invention, the contrast formula is used to accurately calculate the brightness and color images after uniform calibration to obtain the contrast of the image; contrast is one of the important indicators for measuring the display effect of the display, which reflects the difference between the bright and dark parts of the image, and is crucial for the display of the layering and three-dimensional sense of the image; clarity evaluation calculates the clarity index of the image by applying the Laplace operator; Laplace operator, as an edge detection operator, can effectively capture the edge information in the image, thereby evaluating the clarity of the image; a clear image can bring a more delicate and realistic visual experience to the user; color consistency analysis calculates the color difference between different areas or pixels in the image to obtain color consistency parameters; color consistency is an important indicator for measuring the uniformity of the color performance of the display, ensuring that the display can maintain a consistent color effect when presenting different colors, avoiding color deviation and distortion; comprehensive statistics and analysis, based on the contrast and clarity indices calculated above The color standard and color consistency parameters are counted and analyzed to obtain comprehensive analysis results; it is helpful to comprehensively evaluate the display effect of the monitor and provide a strong basis for subsequent optimization and improvement; to improve the display effect, through precise contrast and clarity calculation, as well as color consistency analysis, it is possible to timely discover the problems and deficiencies in the display effect of the monitor, which can significantly improve the display effect of the monitor and bring users a better visual experience; to ensure product quality, comprehensive display effect analysis provides strong support for quality control in the production process; through regular or irregular analysis, problems in the production process can be discovered in a timely manner to ensure that each monitor can meet high quality standards; to enhance user satisfaction, high-quality display effect is one of the key factors to improve user satisfaction; through precise display effect analysis and optimization, it can ensure that the monitor meets user expectations in terms of color reproduction, contrast, clarity, etc., thereby enhancing user satisfaction and loyalty to the product.

[0097] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high-precision calibration method in TFT display production, characterized in that: include: Establish particle swarms for each type of TFT display respectively, and obtain the optimal intermediate value of brightness and contrast of each type of TFT display by simulating the movement of particles in the search space, that is, the initial value, wherein each particle in the particle swarm represents the brightness and contrast of a group of TFT displays; determine the particles representing the brightness and contrast of each group of TFT displays, calculate the fitness of the brightness and contrast of each particle, determine the optimal fitness of each particle and the global optimal fitness according to the fitness of each pair of particles, and update the speed and position of each particle to obtain the optimal intermediate value of brightness and contrast of each type of TFT display, that is, the initial value; According to the initial values, gene libraries of various models of TFT displays are respectively established, and the brightness and contrast of various models of TFT displays are adjusted and iteratively screened through selection, crossover and mutation to obtain the brightness and contrast of various models of TFT displays that reach the predetermined standard values, that is, the calibrated brightness and contrast of various models of TFT displays, wherein the gene codes in the gene library represent different brightness and contrast of various models of TFT displays; According to the calibrated brightness and contrast of each model of TFT display, the color of the TFT display is adjusted using a deep learning color correction model to obtain color parameters that reach a predetermined standard value, i.e., the calibrated color parameters; According to the calibrated color parameters, the Gamma value is adjusted according to the brightness and color distribution of the displayed content using an adaptive neural network model to obtain a calibrated Gamma value; According to the calibrated Gamma value, the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen of each model to obtain uniform calibrated brightness and color; each area of ​​the TFT display screen is regarded as each node, the potential energy function formula of each node is determined, and the objective function of the Markov random field algorithm is constructed according to the potential energy function of each node. The objective function of the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen to obtain uniform calibrated brightness and color; According to the uniformly calibrated brightness and color, the image analysis algorithm is used to analyze the display effects of various types of TFT displays to obtain analysis results; According to the analysis result, a comparison is performed with a predetermined standard value to obtain a comparison result. If the comparison result exceeds or reaches a threshold, recalibration is performed. If the comparison result does not exceed the threshold, the calibration is completed.

2. The high-precision calibration method in TFT display production according to claim 1, characterized in that: A particle swarm of each type of TFT display is established respectively, and the optimal intermediate value of the brightness and contrast of each type of TFT display, i.e., the initial value, is obtained by simulating the movement of particles in the search space, wherein each particle in the particle swarm represents the brightness and contrast of a group of TFT displays, including: Determine particles representing the brightness and contrast of each group of TFT displays, each particle representing the brightness and contrast of a group of TFT displays; use Calculate the fitness of each particle's brightness and contrast, where is the target brightness, is the target contrast, is brightness, is the contrast; According to the fitness of each pair of particles, the optimal fitness of each particle and the global optimal fitness are determined; According to the optimal fitness of each particle and the global optimal fitness, use , The speed and position of each particle are updated to obtain the optimal intermediate value of brightness and contrast of each type of TFT display, that is, the initial value, where: and are the velocity and position of particle i, is the inertia weight, and are learning factors, which control the speed at which particles learn to the optimal solution and the global optimal solution respectively. is a random number between [0,1], is the optimal position of particle i, and gbest is the global optimal position.

3. The high-precision calibration method in TFT display production according to claim 1, characterized in that: According to the initial values, gene libraries of various types of TFT displays are established respectively, and the brightness and contrast of various types of TFT displays are adjusted and iteratively screened through selection, crossover and mutation to obtain the brightness and contrast of the display reaching the predetermined standard value, that is, the calibrated brightness and contrast of various types of TFT displays, wherein the gene codes in the gene library represent different brightness and contrast of various types of TFT displays, including: According to the initial values, the gene library of each type of TFT display is established respectively, and the gene codes in the gene library represent different brightness and contrast; According to the fitness of each pair of particles, determine the parent particle; According to the particles of the parent generation, two parent generation particles are randomly selected to generate two child generation particles in a double-point crossover manner to obtain a number of child generation particles; According to the offspring particles, the gene positions are randomly changed to obtain the mutated particles; According to the mutated particles, fitness evaluation is performed, and particles whose fitness reaches or exceeds the threshold are selected to enter the next generation; The above steps are iterated until the brightness and contrast of each model of TFT display reaching the predetermined standard value, ie, the calibrated brightness and contrast, are obtained.

4. The high-precision calibration method in TFT display production according to claim 1, characterized in that: According to the calibrated brightness and contrast of each model of TFT display, the color of the TFT display is adjusted using a deep learning color correction model to obtain color parameters that reach a predetermined standard value, i.e., calibrated color parameters, including: Acquiring color parameters of the TFT display and color parameters of predetermined standard values; Build deep learning models using The difference between the color parameters of the TFT display and the color parameters of the predetermined standard values ​​is calculated, wherein: z is the loss value, is the sample size, It is The target color parameters for samples, is the color parameter predicted by the model; According to the difference between the color parameters of the TFT display and the color parameters of the predetermined standard values, the deep learning model is optimized using a back propagation algorithm to obtain an optimized deep learning model; According to the color parameters of the TFT display and the color parameters of the predetermined standard values, the color parameters of the TFT display are calibrated using a deep learning model to obtain color parameters that reach the predetermined standard values, that is, calibrated color parameters.

5. The high-precision calibration method in TFT display production according to claim 1, characterized in that: According to the calibrated color parameters, the Gamma value is adjusted according to the brightness and color distribution of the displayed content using an adaptive neural network model to obtain a calibrated Gamma value, including: Obtaining the Gamma value of the TFT display and the Gamma value of a predetermined standard value; According to the Gamma value of the TFT display and the Gamma value of the predetermined standard value, use , Gamma value adjustment and calibration are performed to obtain a calibrated Gamma value, where: is the first input feature Quantity, It is Input the components to The weight of a neuron, is the bias term, It is The net input to each neuron is is the activation function, is the output of the f-th neuron.

6. The high-precision calibration method in TFT display production according to claim 1, characterized in that: According to the calibrated Gamma value, the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​each model of TFT display screen to obtain uniform calibrated brightness and color, including: Obtaining brightness and color characteristics of each area of ​​the TFT display screen, wherein the characteristics include average brightness, brightness standard deviation, color mean, and color variance; Treat each area of ​​the TFT display screen as each node and determine the neighborhood relationship between the nodes; According to the neighborhood relationship between nodes, the potential energy function formula of each node is determined, that is, To obtain the potential energy function of each node, the potential energy function includes a data term and a smoothing term, where: is the total potential energy function, is the data item potential energy function of node g, is the smooth potential function between nodes g and h, N is the set of neighborhood relations, is the state of node g, where g and h are two adjacent nodes respectively; According to the potential energy function of each node, the objective function of the Markov random field algorithm is constructed, that is, ,in, is the objective function, Represents the search for the node state configuration that minimizes the total potential energy ; The objective function of the Markov random field algorithm is used to adjust the brightness and color uniformity of each area of ​​the TFT display screen to obtain uniformly calibrated brightness and color.

7. The high-precision calibration method in TFT display production according to claim 1, characterized in that: According to the uniformly calibrated brightness and color, the image analysis algorithm is used to analyze the display effect of the TFT display to obtain the analysis results, including: Based on the uniformly calibrated brightness and color image, use Calculate the contrast of the image, where is the contrast, is the brightness change, is brightness; Based on the uniformly calibrated brightness and color image, use Calculate the image clarity index, where is the image function, is the clarity index, Indicates the change in the curvature of the image in the horizontal direction. It reflects the change of curvature of the image in the vertical direction; Based on the uniformly calibrated brightness and color image, use Calculate the color difference between different regions or pixels in the image to obtain the color consistency parameter, where: is the color consistency parameter, and are the red, green, and blue component values ​​of the two pixels respectively; According to the image contrast, clarity index, and color consistency parameters, statistics are calculated to obtain analysis results.

Images