A method and system for detecting and correcting color difference of an LED liquid crystal display screen

Abstract

The application discloses a kind of LED liquid crystal display screen bright chromatic aberration detection correction method and system, belong to liquid crystal display technical field, obtain the initial luminance value and color value of each display unit of display screen;Bright chromatic aberration reference model is constructed based on convolutional neural network, the luminance deviation ΔL and color deviation ΔC between calculation and standard value;Bright chromatic aberration error matrix W is constructed;According to error matrix and combined with adaptive threshold algorithm, identify bright chromatic aberration abnormal area and its corresponding coordinates;Based on deviation score result, introduce gradient regression model to generate luminance gain coefficient and color adjustment coefficient, construct correction parameter set P;Parameter set is issued to control chip, executes dynamic image signal modulation and local frame level refresh, realizes point by point bright chromatic aberration correction;The method has the advantages of high precision, high automation, strong adaptive capacity etc., is suitable for the bright chromatic aberration uniformity promotion in a variety of LED liquid crystal display screen scene.

Description

Technical Field

[0001] This invention relates to the field of liquid crystal display technology, and specifically to a method and system for detecting and correcting brightness and color difference in LED liquid crystal displays. Background Technology

[0002] With the continuous development of display technology, LED LCD screens, due to their advantages such as high brightness, low power consumption, and vibrant colors, are widely used in advertising media, traffic information dissemination, security monitoring, and command centers. To ensure consistent display effects and accurate image reproduction, LED LCD screens typically undergo brightness and color uniformity calibration before leaving the factory. However, due to manufacturing errors between modules, varying degrees of aging of LED chips, and differences in drive circuit response, differences in brightness and color exist in different areas, easily leading to "brightness-color difference," a phenomenon where local brightness and color are inconsistent, thus affecting the user's viewing experience.

[0003] Existing methods for luminance and color difference correction mostly rely on manual visual inspection or image sensor-based data acquisition and analysis. However, these methods suffer from low accuracy, inefficiency, and high cost, and are difficult to implement for fine-grained correction in large-size display scenarios. Therefore, there is an urgent need for an automated, high-precision, and low-cost method for detecting and correcting luminance and color difference in LED LCD displays to improve display consistency and overall display quality. This has become a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for detecting and correcting brightness and color difference in LED liquid crystal displays, so as to overcome the shortcomings in the prior art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting and correcting brightness and color difference in an LED liquid crystal display screen, comprising:

[0006] S100: Obtain the initial brightness and color values ​​of each display unit on the LED LCD screen;

[0007] S200. Based on the standard brightness and color reference model, calculate the brightness deviation ΔL and color deviation ΔC between each display unit and the standard value;

[0008] S300. Based on ΔL and ΔC, construct the luminance-color difference error matrix W;

[0009] S400. Analyze the error matrix W according to the preset threshold, identify abnormal areas of brightness and color difference and determine the position coordinates of the corresponding display unit;

[0010] S500: Generate a corresponding set of correction parameters P based on the degree of deviation in the abnormal areas of brightness and color difference, including brightness gain coefficient and color adjustment coefficient;

[0011] S600: Apply the correction parameter set P to the corresponding display unit and perform brightness and color difference correction operation;

[0012] S700, Repeat steps S100 to S600 until the brightness and color deviation of all display units are less than the preset threshold.

[0013] Preferably, S200 specifically includes:

[0014] S201. Construct a brightness and color difference reference model based on a convolutional neural network. The input of the model is the original RGB data of each display unit and the ambient lighting parameters. The output is the target brightness value Lref and the target color value Cref.

[0015] S202. Standardize the current brightness Lcur and color value Ccur of each display unit that is actually collected;

[0016] S203. The weighted difference algorithm is used to calculate the brightness deviation ΔL and color deviation ΔC of each display unit.

[0017] Preferably, S300 specifically includes:

[0018] S301. Divide the display screen into a grid according to the pixel array, establish a two-dimensional coordinate system that corresponds one-to-one with the physical location, and map the ΔL and ΔC values ​​of each display unit respectively.

[0019] S302. Principal component analysis is used to compress the features of ΔL and ΔC, extracting the dominant components of brightness and color deviation, and retaining high variance features.

[0020] S303. Embed the compressed feature vector into a two-dimensional error tensor structure to construct a luminance-color difference error matrix W, where each element of the matrix is ​​generated by combining the principal deviation components of the corresponding pixel.

[0021] Preferably, S400 specifically includes:

[0022] S401. Perform multi-threshold segmentation on the brightness and color difference error matrix W, and use the adaptive Otsu algorithm to dynamically determine the optimal boundary threshold of the joint deviation of ΔL and ΔC in order to obtain the preliminary abnormal region mask.

[0023] S402. Based on regional connectivity analysis, perform morphological filtering on the mask of abnormal regions to extract stable bright-color difference abnormal contours.

[0024] S403. Map the abnormal contour back to the original display unit grid, and combine the matrix index and coordinate mapping function to determine the two-dimensional position coordinates (i,j) of the corresponding display unit.

[0025] Preferably, S500 specifically includes:

[0026] S501. For each region with abnormal brightness and color, calculate its comprehensive deviation score R based on the amplitude and distribution density of its corresponding ΔL and ΔC, where R is formed by weighted fusion of brightness deviation intensity, color shift angle and region area.

[0027] S502. An adaptive gradient regression model is introduced based on the score R to fit and generate the brightness gain coefficient GL and color adjustment coefficient GC of the corresponding display unit. The model learns the optimal correction strategy under different deviation modes through training samples.

[0028] S503. Combine GL and GC into a correction parameter pair P(i,j) = {GL(i,j), GC(i,j)} and propagate the interpolation within the anomalous region;

[0029] S504. Construct a correction parameter set P, which contains parameter pairs P(i,j) of all display units within the abnormal region, and binds them with a unique region identifier ID and coordinate index.

[0030] Preferably, S600 specifically includes:

[0031] S601. The calibration parameter set P is sent to the LED display control module through the drive control interface, and point-by-point mapping and binding is performed according to the display unit coordinates (i,j);

[0032] S602. Set up a real-time correction register channel in the display control chip, and dynamically modulate the input image signal according to the brightness gain coefficient GL and color adjustment coefficient GC in P(i,j), including brightness amplitude reconstruction and RGB color matrix rotation transformation.

[0033] S603. Simultaneously perform local redrawing operations and refresh the updated area at the frame level;

[0034] S604: Monitor the residual change between the output of each frame of the image after correction and the target brightness and color model. If the residual exceeds the tolerance range, automatically return to S500 for parameter adaptive iterative optimization.

[0035] The present invention also provides an LED liquid crystal display screen brightness and color difference detection and correction system, comprising:

[0036] The image acquisition module acquires the initial brightness and color values ​​of each display unit on the LED LCD screen.

[0037] The deviation calculation module calculates the brightness deviation ΔL and color deviation ΔC between each display unit and the standard value based on the standard brightness and color reference model.

[0038] The error modeling module constructs the luminance-color difference error matrix W based on ΔL and ΔC.

[0039] The anomaly detection module analyzes the error matrix W according to a preset threshold, identifies abnormal areas of brightness and color difference, and determines the position coordinates of the corresponding display unit.

[0040] The parameter generation module generates a corresponding set of correction parameters P, including brightness gain coefficient and color adjustment coefficient, based on the degree of deviation in the abnormal areas of brightness and color difference.

[0041] The image correction module applies the correction parameter set P to the corresponding display unit and performs brightness and color difference correction operation.

[0042] The control and feedback module is repeatedly optimized until the brightness and color deviation of all display units are less than the preset threshold.

[0043] The technical effects and advantages provided by the present invention in the above technical solution are as follows:

[0044] 1. This invention introduces a brightness and color difference reference model based on a convolutional neural network, combined with principal component extraction of the deviation vector and error matrix modeling, to achieve high-precision identification and spatial mapping of brightness and color deviations in LED LCD displays. Compared with traditional solutions relying on fixed templates or manual inspection, this invention has stronger adaptability and stability, and can dynamically respond to complex changes in different screen types, display content, and lighting environments, significantly improving detection accuracy and anomaly localization accuracy.

[0045] 2. This invention constructs a closed-loop self-calibration mechanism for parameter generation and application. It adaptively outputs brightness gain coefficients and color adjustment coefficients through a gradient regression model, and implements dynamic image modulation and local refresh in the control chip. Supplemented by error residual feedback control, it effectively avoids over-correction and visual fragmentation. The overall technical approach boasts advantages such as high automation, strong scalability, and good visual consistency, making it suitable for high-end display applications with extremely high requirements for image uniformity and color accuracy. 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 embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0047] Figure 1 This is a mind map of the method of the present invention.

[0048] Figure 2 This is a mind map of the system modules of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Example 1, please refer to Figure 1 As shown in this embodiment, a method for detecting and correcting brightness and color difference in an LED liquid crystal display screen includes:

[0051] S100: Obtain the initial brightness and color values ​​of each display unit on the LED LCD screen;

[0052] S200. Based on the standard brightness and color reference model, calculate the brightness deviation ΔL and color deviation ΔC between each display unit and the standard value;

[0053] S300. Based on ΔL and ΔC, construct the luminance-color difference error matrix W;

[0054] S400. Analyze the error matrix W according to the preset threshold, identify abnormal areas of brightness and color difference and determine the position coordinates of the corresponding display unit;

[0055] S500: Generate a corresponding set of correction parameters P based on the degree of deviation in the abnormal areas of brightness and color difference, including brightness gain coefficient and color adjustment coefficient;

[0056] S600: Apply the correction parameter set P to the corresponding display unit and perform brightness and color difference correction operation;

[0057] S700, Repeat steps S100 to S600 until the brightness and color deviation of all display units are less than the preset threshold.

[0058] Specifically, S100 includes the following steps:

[0059] A high-resolution imaging device is deployed in front of the LED LCD screen, preferably using an industrial camera module with wide dynamic range (HDR), combined with a uniformly distributed illumination reference board to ensure minimal ambient light interference during the acquisition process. The imaging device is fixed to the screen directly in front of it using a stable bracket, with the acquisition angle perpendicular to the screen or controlled within a range of less than 5 degrees to improve the geometric consistency of the image.

[0060] The control display screen plays a standard test image sequence, including pure white, pure red, pure green, pure blue, grayscale bars, and color gradient stripes, to stimulate the brightness and color response of the LED units under typical operating conditions. During playback, a constant brightness level and refresh rate are maintained to avoid image fluctuations affecting data acquisition.

[0061] The camera captures images frame by frame, and the original images undergo denoising, distortion correction, and gamma inverse correction. Bilateral filtering or median filtering is used to remove small areas of noise, and camera calibration parameters are used for image correction to ensure strict alignment between the image data and the physical layout of the display screen.

[0062] The processed image is divided into pixel grid regions corresponding to each display unit (e.g., each module corresponds to an NxN pixel block). Pixel averaging is performed on each region to extract representative luminance values ​​Lcur and color values ​​Ccur. The luminance value can be calculated based on the Y channel (e.g., converted from RGB to YCbCr space), and the color value is represented using CIE Lab or HSV space to enhance response to areas sensitive to the human eye.

[0063] Lcur and Ccur are bound to the two-dimensional coordinates (i,j) of the display unit to form a complete data structure D(i,j) = {Lcur(i,j), Ccur(i,j)}, and stored in the initialization data matrix to provide basic input for the deviation calculation in the subsequent step S200.

[0064] Specifically, S200 includes the following sub-steps:

[0065] A deep learning model is established as a luminance-color difference reference standard generator, employing a multi-layer convolutional neural network (CNN) structure. The input layer receives the three-channel RGB values ​​of each display unit and ambient lighting parameters (such as light intensity, direction, and color temperature). The intermediate layers extract local color texture and luminance gradient features through multi-scale convolutional kernels, and fuse BatchNorm and ReLU activation to enhance the model's robustness.

[0066] The model is trained on a large amount of display sample data and can output target brightness value Lref and target color value Cref, representing the brightness and color performance under ideal display conditions, respectively. The output values ​​physically reflect the screen's factory standard or user-defined benchmark and have cross-screen generalization capabilities.

[0067] For the actual collected Lcur and Ccur values, normalization and illumination compensation processing are performed. The brightness value Lcur is obtained using the maximum-minimum normalization method, and linear illumination inversion is performed in conjunction with the ambient light intensity to remove the influence of overall exposure differences on the brightness of local units.

[0068] The color value Ccur is converted to Lab or HSV color space, channel normalized, and median smoothing is used to suppress edge pixel abruptness. This processing step ensures that the deviation values ​​in subsequent calculations have cross-regional consistency and contrast.

[0069] For each display unit (i,j), calculate its brightness deviation according to the formula. Color deviation A region weighting function W(i,j) is introduced, which is adaptively generated based on pixel neighborhood consistency and local gradient change rate, and is used to adjust the weights affected by deviations.

[0070] In particular, the color deviation ΔC is calculated using a vector difference method, which comprehensively considers the difference in hue angle and saturation, thus avoiding the offset distortion problem of traditional Euclidean distance in non-uniform color space.

[0071] Finally, ΔL and ΔC are output as error information for subsequent steps, which has high resolution, high reliability and interpretability.

[0072] Specifically, S300 includes the following sub-steps:

[0073] The LED LCD screen is divided into an M×N pixel array grid, with each grid cell (i,j) corresponding to an actual display pixel cell. Based on the previously collected and calculated ΔL(i,j) and ΔC(i,j) data, a two-dimensional coordinate system is constructed to map the position of each display cell to its brightness and color difference, forming a brightness deviation field ΔLmap and a color deviation field ΔCmap.

[0074] This spatial mapping ensures that all deviation information can be mapped to the actual physical display area, providing a basic positional reference for subsequent matrix construction.

[0075] Principal component analysis (PCA) is performed on the deviation vector set composed of ΔLmap and ΔCmap to extract the eigenvalues ​​that contribute the most to the variance. This step is used to remove redundant information, compress high-dimensional deviation data, and preserve the changing trends that are most sensitive to anomalous regions of brightness and color difference.

[0076] During compression, the first k principal components (usually k=1 or 2) are retained to ensure that more than 95% of the original bias information is not lost, while significantly improving data processing efficiency.

[0077] The compressed principal feature components are remapped back to the original 2D coordinate grid and embedded into a 2D tensor structure W. W is an M×N matrix, where each element W(i,j) represents the synthesized value of the principal bias of the display unit, in the form of: ; where α and β are the weighting coefficients of the principal components of brightness and color, which can be customized according to actual display preferences or visual perception models.

[0078] The error matrix W can be viewed as a numerical "deviation map" used to describe the overall deviation distribution of the entire display screen in terms of brightness and color, providing a quantitative basis for subsequent abnormal area identification and correction strategy formulation.

[0079] Specifically, S400 includes the following sub-steps:

[0080] For the constructed luminance-color difference error matrix W, the combined deviation value W(i,j) of ΔL and ΔC is extracted as the basic segmentation criterion, and an adaptive multi-threshold segmentation method is used for region division. Specifically, based on the improved Otsu algorithm, a set of optimal grayscale thresholds is dynamically determined. , where n is the total number of grayscale thresholds, maximizes the inter-class variance and adaptively adjusts the number of segmentation layers.

[0081] This method can accurately classify normal areas and multiple abnormal levels of brightness and color difference based on the actual variance of the deviation value in the full-screen distribution, avoiding misjudgment or omission caused by uneven regional distribution due to the fixed threshold method.

[0082] The output is a binary mask image M, where M(i,j) = 1 indicates an abnormal region and M(i,j) = 0 indicates a normal region, forming the boundary outline of the abnormal region with initial light and color differences.

[0083] A region connectivity analysis is performed on the initial mask image M, using the 8-neighborhood connectivity rule to identify all closed or nearly closed bright color difference anomalous blocks. To enhance the stability of region boundaries and suppress noise interference, morphological opening and closing operations (opening operation for noise reduction, closing operation for hole filling) are further performed on the mask image to remove spurious differences and refine the contour shape.

[0084] After filtering, extract the contours of all stable anomaly regions. k represents the total number of abnormal region contours; the set of boundary points and regional characteristic parameters (area, shape factor, etc.) are recorded to provide structural input for localization and correction.

[0085] Each abnormal contour region Ri is reverse-mapped into the original display unit grid, and precise positioning is achieved using matrix indexing and a two-dimensional coordinate mapping function to generate the coordinate set of the corresponding unit. .

[0086] The mapping process maintains the alignment of the error matrix W with the physical pixel array structure and introduces position offset corrections (such as camera distortion compensation parameters) when necessary, ensuring that the identified abnormal areas have practically operable positioning accuracy on the display screen.

[0087] The output is a set of coordinates for the abnormal display unit and an index for the contour region, which can be used to establish the region label and coordinate binding relationship in the subsequent correction parameter generation step.

[0088] Specifically, S500 includes the following sub-steps:

[0089] For each identified area with abnormal brightness or color, the ΔL and ΔC values ​​of its internal display unit are extracted, and the comprehensive deviation score R is calculated based on the following three dimensions:

[0090] Brightness deviation intensity The root mean square (RMS) of the ΔL value within the region reflects the degree of brightness non-uniformity.

[0091] Color shift angle ΔC is the vector deviation angle in the Lab or HSV color space, used to quantify the hue direction error;

[0092] Regional area weight The number of display units contained in the abnormal area reflects its impact on the entire screen.

[0093] score ,in These are empirical weighting coefficients, adjusted and optimized based on actual scenarios.

[0094] An adaptive gradient regression model (such as gradient boosting regression tree GBRT or lightweight neural network model) is introduced, with the score R as the input feature, and the fitted output targets are the brightness gain coefficient G_L and the color adjustment coefficient GC.

[0095] The training data comes from a historical calibration sample library. The sample comparisons with the ΔL and ΔC deviation types and the manually or simulated optimized correction parameters form supervised learning input-output pairs. This model has nonlinear fitting capabilities, can adapt to different brightness and color difference patterns, and output refined parameter prediction results.

[0096] Each display unit within the abnormal area can independently obtain the corresponding parameter pair: .

[0097] To eliminate display discontinuities caused by abrupt changes in region boundary correction parameters, bilinear interpolation or boundary weighted smoothing interpolation algorithms are employed to continuously interpolate and diffuse GL and GC within the anomalous region. The interpolation algorithm considers the deviation gradient variation trend from the region center to the edge, ensuring a natural spatial transition of the correction parameters.

[0098] In addition, a "buffer expansion" is performed on the edges of adjacent non-abnormal areas to construct a transition layer, further reducing the visual discontinuity caused by the correction.

[0099] Finally, all parameter pairs P(i,j) within the abnormal regions are packaged and encapsulated according to region number to form a structured correction parameter set P, which includes:

[0100] Unique regional identifier (ID);

[0101] The coordinates (i,j) of each display unit within the area;

[0102] The corresponding brightness gain coefficient GL and color adjustment coefficient GC;

[0103] Regional bias score R and model fit status.

[0104] The parameter set P is scalable, callable, and traceable, making it easy for subsequent driver modules to connect and perform correction control in real time.

[0105] Specifically, S600 includes the following sub-steps:

[0106] The pre-constructed calibration parameter set P is sent to the LED display control module via the main control system's drive control interface (such as SPI, I2C, or FPGA on-chip bus). The parameter data packet contains fields such as display unit coordinates (i,j), brightness gain coefficient GL(i,j), and color adjustment coefficient GC(i,j).

[0107] Based on the coordinate index, the control module precisely binds the parameters to the corresponding drive channel or control register, ensuring that each pixel receives a unique corresponding correction instruction, thus realizing a high-precision control strategy of "point-to-point mapping and point-to-point correction".

[0108] The LED display control chip incorporates a real-time calibration register channel for inserting calibration modules into the image signal processing chain. The specific implementation is as follows:

[0109] Luminance gain modulation is performed on the original image input signal for each pixel, and the luminance amplitude is dynamically reconstructed according to GL(i,j);

[0110] Perform color matrix rotation transformation or gamma space mapping correction on the RGB channel values, and fine-tune the hue direction and saturation according to GC(i,j).

[0111] The above processing runs online at the front end of the display control frame buffer, completing vector correction of all pixel-level images before each frame refresh, ensuring that the correction effect is synchronized with the image in real time.

[0112] To improve system response efficiency, the control module supports a local area-level redraw refresh strategy. Redraw operations are only performed on abnormal areas where correction parameters have been applied, triggering data updates in the corresponding areas of the frame buffer, thus avoiding resource waste and display jitter caused by full-screen refreshes.

[0113] The refresh operation is synchronized with the display refresh rate (such as 60Hz or 120Hz) to ensure that there is no lag in display after the correction parameters take effect, thus improving the user's visual consistency experience.

[0114] After each frame of image is output, the system calls the image feedback module or an external image acquisition device to perform a residual comparison between the actual brightness and color output of the current frame and the target reference model (Lref, Cref). If ΔL′ or ΔC′ in some areas is detected to exceed the tolerance range (e.g., ±5% or ΔE>3), it is considered that the correction effect is insufficient.

[0115] At this point, the system automatically triggers the closed-loop mechanism, returns to step S500, re-evaluates the score R, and constructs a new round of correction model based on historical residuals to achieve adaptive iterative optimization and continuously improve correction accuracy.

[0116] Specifically, S700 includes the following sub-steps:

[0117] The system is designed as a closed-loop brightness and color difference detection-correction feedback framework. After the first execution of S100 to S600, the corrected display state is used as input again and sent back to step S100 to re-collect the brightness value Lcur' and color value Ccur' of each display unit for the second round of error evaluation.

[0118] This iterative mechanism ensures that each round of correction is based on the latest state, avoiding secondary deviations caused by insufficient or excessive initial correction.

[0119] After each iteration, steps S200–S300 are re-executed to calculate new ΔL′ and ΔC′ values ​​and reconstruct the error matrix W′. Error aggregation analysis is performed on the full-screen display unit to extract the following two metrics:

[0120] The maximum deviation value is max(ΔL′, ΔC′).

[0121] The average deviation is mean(ΔL′, ΔC′).

[0122] The system compares these two indicators with preset thresholds, such as ΔL threshold = 3% and ΔC threshold (color difference ΔE) = 2.5, to determine whether the current correction result has reached a stable state.

[0123] If the deviation of any unit still exceeds the threshold, the system automatically enters the next iteration; if all units ΔL′ and ΔC′ are less than the corresponding threshold, or the error change rate of two consecutive iterations is lower than the set convergence threshold (e.g., <0.2%), it is judged as correction convergence and the loop is terminated.

[0124] The system supports setting the maximum number of iterations Nmax (e.g., 3–5 times) to prevent infinite loops in abnormal situations, thereby improving operational safety and stability.

[0125] Upon termination of the iteration, the system outputs the final calibration parameter set Pfinal, the error matrix Wfinal, and error change logs for each round, and permanently stores the calibration parameters in the controller Flash as the system's default calibration state. Optionally, a calibration heatmap and convergence trend curve are output for visual quality assessment.

[0126] The system will automatically switch the display control status to "lock mode" to stop subsequent image signal modulation intervention and ensure the consistency and stability of the final display status.

[0127] Example 2, please refer to Figure 2 As shown in this embodiment, an LED LCD screen brightness and color difference detection and correction system includes:

[0128] The image acquisition module acquires the initial brightness and color values ​​of each display unit on the LED LCD screen.

[0129] The deviation calculation module calculates the brightness deviation ΔL and color deviation ΔC between each display unit and the standard value based on the standard brightness and color reference model.

[0130] The error modeling module constructs the luminance-color difference error matrix W based on ΔL and ΔC.

[0131] The anomaly detection module analyzes the error matrix W according to a preset threshold, identifies abnormal areas of brightness and color difference, and determines the position coordinates of the corresponding display unit.

[0132] The parameter generation module generates a corresponding set of correction parameters P, including brightness gain coefficient and color adjustment coefficient, based on the degree of deviation in the abnormal areas of brightness and color difference.

[0133] The image correction module applies the correction parameter set P to the corresponding display unit and performs brightness and color difference correction operation.

[0134] The control and feedback module is repeatedly optimized until the brightness and color deviation of all display units are less than the preset threshold.

[0135] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A LED liquid crystal display screen bright color difference detection correction method, characterized in that: The method comprises the following steps: S100, acquiring initial brightness values and color values of each display unit on an LED liquid crystal display screen; S200, calculating brightness deviation ΔL and color deviation ΔC between each display unit and a standard value according to a standard brightness and color reference model; S200 specifically comprises: S201, constructing a brightness and color difference reference model based on a convolutional neural network, the model input being original RGB data and ambient light parameters of each display unit, and the output being target brightness value Lref and target color value Cref; S202, performing standardization processing on current brightness Lcur and color value Ccur of each display unit actually collected; S203, calculating brightness deviation ΔL and color deviation ΔC of each display unit by using a weighted difference algorithm; S300, constructing a brightness and color difference error matrix W according to ΔL and ΔC; S400, analyzing the error matrix W according to a preset threshold value, identifying a brightness and color difference abnormal area and determining position coordinates of the display unit corresponding to the area; S500, generating a corresponding correction parameter set P including a brightness gain coefficient and a color adjustment coefficient according to the deviation degree of the brightness and color difference abnormal area; S500 specifically comprises: S501, for each brightness and color difference abnormal area, calculating a comprehensive deviation score R according to the amplitude and distribution density of ΔL and ΔC corresponding to the area, wherein R is formed by weighted fusion of brightness deviation intensity, color shift angle and area size; S502, introducing an adaptive gradient regression model based on the score R to respectively fit and generate a brightness gain coefficient GL and a color adjustment coefficient GC of the corresponding display unit, the model learning an optimal correction strategy under different deviation modes through training samples; S503, combining GL and GC into a correction parameter pair P(i,j) = {GL(i,j), GC(i,j)} and interpolating and propagating in the abnormal area; S504, constructing the correction parameter set P containing the parameter pair P(i,j) of the display unit in all abnormal areas and binding a unique area identifier ID and a coordinate index; S600, applying the correction parameter set P to the corresponding display unit to perform a brightness and color difference correction operation; S700, repeating steps S100 to S600 until the brightness and color deviation of all display units are less than the preset threshold value.

2. The LED liquid crystal display screen bright chromatic aberration detection and correction method according to claim 1, characterized in that: S300 specifically comprises: S301, grid dividing the display screen according to a pixel array, establishing a two-dimensional coordinate system corresponding to physical positions, and respectively mapping ΔL and ΔC values of each display unit; S302, performing principal component analysis on ΔL and ΔC to compress features, extract dominant components of brightness and color deviation, and retain high-variance features; S303, embedding the compressed feature vectors into a two-dimensional error tensor structure to construct a brightness and color difference error matrix W, wherein each element of the matrix is generated by combining the principal deviation components of the corresponding pixel points.

3. The LED liquid crystal display screen bright chromatic aberration detection and correction method according to claim 1, characterized in that: S400 specifically comprises: S401, performing multi-threshold segmentation processing on the brightness and color difference error matrix W, and dynamically determining an optimal boundary threshold value of the joint deviation of ΔL and ΔC by using an adaptive Otsu algorithm to obtain a preliminary abnormal area mask; S402, morphological filtering is performed on the abnormal region mask based on region connectivity analysis to extract a stable bright color difference abnormal profile; S403, the abnormal profile is mapped back to the original display unit grid to determine the two-dimensional position coordinates (i, j) of the corresponding display unit in combination with the matrix index and coordinate mapping function.

4. The LED liquid crystal display screen bright chromatic aberration detection and correction method according to claim 1, characterized in that: The S600 specifically includes: S601, the correction parameter set P is issued to the LED display screen control module through the drive control interface, and is mapped and bound point by point according to the display unit coordinates (i, j); S602, a real-time correction register channel is set in the display control chip, and the input image signal is dynamically modulated according to the luminance gain coefficient GL and the color adjustment coefficient GC in P (i, j), including luminance amplitude reconstruction and RGB color matrix rotation transformation; S603, a local redrawing operation is synchronously performed to refresh the update area at the frame level; S604, the residual error change between each frame of image output after correction and the target luminance and color model is monitored, and if the residual error exceeds the tolerance range, the parameter adaptive iterative optimization is automatically returned to S500.

5. A system for detecting and correcting the brightness and color difference of an LED liquid crystal display screen, which is used to implement the method for detecting and correcting the brightness and color difference of an LED liquid crystal display screen according to any one of claims 1-4, characterized in that: It includes: An image acquisition module acquires initial luminance and color values of each display unit on an LED liquid crystal display screen; A deviation calculation module calculates luminance deviation ΔL and color deviation ΔC between each display unit and a standard value according to a standard luminance and color reference model; An error modeling module constructs a luminance and color difference error matrix W according to ΔL and ΔC; An abnormality detection module analyzes the error matrix W according to a preset threshold, identifies a luminance and color difference abnormal region, and determines the position coordinates of the corresponding display unit; A parameter generation module generates a corresponding correction parameter set P including a luminance gain coefficient and a color adjustment coefficient according to the deviation degree of the luminance and color difference abnormal region; An image correction module applies the correction parameter set P to the corresponding display unit to perform a luminance and color difference correction operation; A control and feedback module repeatedly optimizes until the luminance and color deviation of all display units is less than the preset threshold.

Images