Image feature matching-based miniled display screen chroma uniformity correction method and device, and display screen

Abstract

This invention provides a method, apparatus, and display screen for color uniformity correction of minimized displays based on image feature matching, relating to the field of display technology. First, it acquires and records cross-unit transmission images of color features, including the diffusion and cross-over effects of color features propagation in light-emitting units. Next, it constructs a closed-loop link for cross-unit transmission of color features, incorporating forward transmission and reverse feedback. Then, it defines a dynamically adaptable transmission rule set. Next, it integrates and generates a multi-dimensional coupling correction signal matrix. Finally, it iteratively transmits the signal matrix within the closed-loop link, driving the light-emitting units to synchronously adjust their color output, and iteratively updates the signal matrix until the color uniformity standard is met. This invention comprehensively considers the transmission of color features between light-emitting units, significantly improving the color uniformity and display quality of minimized displays.

Description

Technical Field

[0001] This invention relates to the field of display technology, and more specifically, to a method, apparatus, and display screen for color uniformity correction of a minimized display screen based on image feature matching. Background Technology

[0002] Miniled displays hold a significant position in the high-end display market due to their advantages such as high contrast, high brightness, and high resolution. However, in practical applications, miniled displays face the problem of color uniformity, which seriously affects display quality and user experience. Uneven color can lead to color patches and brightness differences on the display, resulting in a degraded image quality.

[0003] Existing methods for color uniformity correction in minimized displays mostly focus on the independent correction of individual light-emitting units (LEDs). This involves measuring the chromaticity parameters of each LED and then adjusting them individually. However, these methods neglect the chromaticity characteristic transmission phenomenon between LEDs. In actual minimized displays, the chromaticity characteristics of LEDs are not only transmitted to directly adjacent LEDs through the gaps between adjacent units, but also further diffuse to non-directly adjacent LEDs. Furthermore, the chromaticity characteristic transmission between multiple units can create a cross-over effect. Existing correction methods do not consider these complex transmission relationships, making it impossible to comprehensively and accurately solve the color uniformity problem and failing to meet the high-quality image requirements of high-end displays. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a method for correcting the color uniformity of a minimized display screen based on image feature matching, the method comprising:

[0005] Acquire cross-unit transmission images of chromaticity features of a Miniled display screen. These cross-unit transmission images record the transmission and diffusion state of chromaticity features of all light-emitting units to non-directly adjacent light-emitting units through the gaps between adjacent units, as well as the superposition effect of cross-transmission of multiple units.

[0006] Based on the chromaticity feature transmission paths of directly adjacent and non-directly adjacent light-emitting units, a closed-loop structure containing forward transmission and reverse feedback is formed by connecting them in series, thus constructing a closed-loop link for chromaticity feature transmission across units.

[0007] Based on the chromaticity feature transmission characteristics under different working scenarios, the chromaticity feature transmission threshold, transmission attenuation coefficient and cross-superposition adaptation method of each light-emitting unit are defined to generate a dynamic adaptation transmission rule set.

[0008] Based on the chromaticity feature cross-unit transmission image, the chromaticity feature cross-unit transmission closed-loop link, and the dynamic adaptation transmission rule set, the differentiated correction parameters of each light-emitting unit in different transmission directions and the cross-unit collaborative correction correlation parameters are integrated to generate a multi-dimensional coupled correction signal matrix.

[0009] In the chromaticity feature cross-unit transmission closed-loop link, the transmission of the multi-dimensional coupling correction signal matrix and the process of driving all light-emitting units to synchronously adjust the chromaticity output according to the multi-dimensional coupling correction signal matrix are executed cyclically. After each adjustment, the chromaticity feature cross-unit transmission image is re-acquired to update the multi-dimensional coupling correction signal matrix, and iterative correction is performed until the preset chromaticity uniformity standard is met, thereby realizing the chromaticity uniformity correction of the Miniled display screen.

[0010] Furthermore, embodiments of the present invention also provide a color uniformity correction device for a minimized display screen based on image feature matching, comprising:

[0011] A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to perform the above-described image feature matching-based chromaticity uniformity correction method for a minimized display screen by executing the machine-executable instructions.

[0012] In another aspect, embodiments of the present invention also provide a display screen, the display screen comprising:

[0013] The display panel has multiple minimized light-emitting units arranged on it;

[0014] The image feature matching-based Miniled display color uniformity correction device is electrically connected to the display panel and is configured to perform the image feature matching-based Miniled display color uniformity correction method described above.

[0015] Based on the above, by acquiring cross-unit transmission images of chromaticity features, the transmission and diffusion states of chromaticity features in all light-emitting units and the cross-transmission superposition effect of multiple units were recorded. Then, a closed-loop link for cross-unit transmission of chromaticity features was constructed, connecting the chromaticity feature transmission paths of directly adjacent and non-directly adjacent light-emitting units, forming a closed-loop structure including forward transmission and reverse feedback. This more realistically simulates the transmission process of chromaticity features in the display screen, making the correction process more consistent with actual conditions. A dynamic adaptation transmission rule set was generated, defining the chromaticity feature transmission threshold, transmission attenuation coefficient, and cross-superposition adaptation method for each light-emitting unit based on different working scenarios. This allows for flexible adjustment of the correction strategy according to actual working conditions, improving the adaptability and accuracy of the correction. A multi-dimensional coupled correction signal matrix was integrated and generated, comprehensively considering the differentiated correction parameters of each light-emitting unit in different transmission directions and the cross-unit collaborative correction correlation parameters, achieving comprehensive and collaborative correction of the display screen's chromaticity. By cyclically executing signal matrix transmission and synchronous adjustment of light-emitting units in a closed-loop link and performing iterative correction, the color uniformity of the display screen can be gradually optimized until it meets the preset standard. This effectively solves the problem that existing methods cannot fully consider the transmission of color characteristics, and significantly improves the display quality and user experience of the Miniled display screen. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the execution flow of the color uniformity correction method for a minimized display screen based on image feature matching provided in an embodiment of the present invention.

[0017] Figure 2 This is a schematic diagram of exemplary hardware and software components of the Miniled display screen color uniformity correction device based on image feature matching provided in an embodiment of the present invention. Detailed Implementation

[0018] Figure 1 This is a flowchart illustrating a method for correcting the color uniformity of a minimized display screen based on image feature matching, provided in one embodiment of the present invention. A detailed description follows.

[0019] Step S110: Acquire the chromaticity feature cross-unit transmission image of the Miniled display screen. The chromaticity feature cross-unit transmission image records the transmission and diffusion state of the chromaticity features of all light-emitting units to non-directly adjacent light-emitting units through the gaps between adjacent units and the superposition effect of multi-unit cross-transmission.

[0020] In this embodiment, a 3840×2160 resolution Miniled display screen, containing a large number of light-emitting units, requires a high-precision spectral imaging device to acquire the cross-unit transmission image of the aforementioned chromaticity features. This high-precision spectral imaging device should have a spectral resolution of at least 1 nm and a spatial resolution of at least 10 μm to ensure clear capture of subtle chromaticity changes in each light-emitting unit and their transmission between adjacent units. During the acquisition process, the Miniled display screen is first placed in a dark room to avoid external light interference. The display screen is then illuminated and displays a completely white image, at which point each light-emitting unit is in a stable working state. The spectral imaging device is fixed 1 meter directly in front of the display screen, with the lens vertically aligned with the center of the display screen, and the entire display screen is imaged using a line-by-line scanning method. During imaging, the device records the spectral data of each pixel within the visible spectrum, including the spectral intensity distribution of the three primary colors: red, green, and blue. For each light-emitting unit, not only are its own chromaticity features recorded, but also the state of its chromaticity features being transmitted to surrounding non-directly adjacent light-emitting units through the physical gaps between adjacent units. For example, the red light emitted by a light-emitting unit located in the center of the display screen will be transmitted to the diagonally indirect light-emitting units through the gaps between it and the four directly adjacent light-emitting units above, below, left, and right. At the same time, the chromaticity characteristics of other adjacent light-emitting units will also be transmitted to this unit, forming a multi-unit cross-transmission superposition effect. After preliminary processing to remove noise interference, the acquired raw image data forms a chromaticity feature cross-unit transmission image containing the chromaticity feature transmission information of all light-emitting units. This chromaticity feature cross-unit transmission image is stored in three-dimensional data form, where two dimensions correspond to the spatial coordinates of the display screen, and the third dimension records the spectral data and transmission state parameters of each coordinate point.

[0021] Step S120: Based on the chromaticity feature transmission paths of directly adjacent and non-directly adjacent light-emitting units, a closed-loop structure containing forward transmission and reverse feedback is formed by connecting them in series, thus constructing a closed-loop link for chromaticity feature transmission across units.

[0022] After acquiring the chroma feature cross-unit transmission image, a closed-loop link needs to be constructed based on the transmission path information recorded therein. First, the definitions of directly adjacent and indirectly adjacent light-emitting units need to be clarified. Directly adjacent units share an edge in the physical layout of the display screen, while indirectly adjacent units share a vertex but not an edge. Then, the chroma feature transmission direction and intensity of each light-emitting unit are extracted from the transmission image to determine the forward transmission path, i.e., the path from one light-emitting unit to its adjacent units. Simultaneously, the reverse feedback path also needs to be considered. When a light-emitting unit receives a transmission feature from another unit, it generates a feedback signal, which is transmitted back to the original light-emitting unit along the reverse direction of the original transmission path. By concatenating the aforementioned forward transmission path and reverse feedback path, a closed-loop structure encompassing all light-emitting units is formed—the chroma feature cross-unit transmission closed-loop link. In this closed-loop link, each light-emitting unit is both a sender and receiver of the transmission feature, and also a generator of the feedback signal, ensuring that the chroma feature can be dynamically transmitted and fed back across the entire display screen.

[0023] Step S121: Perform luminescent unit and gap region segmentation processing on the chromaticity feature cross-unit transmission image, separate the body region of each luminescent unit, the gap region between adjacent units, and the indirect transmission influence region of non-directly adjacent units, and record the chromaticity feature transmission information associated with the region in the metadata of each region generated after the segmentation processing.

[0024] In this embodiment, a deep learning-based image segmentation algorithm is used to process the chromaticity feature transmission image across units. First, a large number of labeled images of a minimized display screen, including the main body region, gap region, and indirect transmission influence region of each luminescent unit, are prepared as training samples to train a segmentation model based on the U-Net architecture. The input to this segmentation model is the three-dimensional data of the chromaticity feature transmission image across units, and the output is the classification result of the region to which each pixel belongs. During the segmentation process, the image is input into the trained segmentation model. The model extracts and analyzes image features to segment the main body region of each luminescent unit. The main body region is the core region of the luminescent unit, with clear boundaries and high spectral intensity. The gap region between adjacent units refers to the physical gap between directly adjacent luminescent units. This region has relatively low spectral intensity and contains transitional information of chromaticity feature transmission. The indirect transmission influence region of non-directly adjacent units refers to the region that is far from the main body region of the luminescent unit but is still indirectly affected by its chromaticity features. This region has weak spectral intensity, and the chromaticity features are somewhat correlated with the main body region. After segmentation, corresponding metadata is generated for each region. The metadata includes the region's location coordinates, boundary information, average spectral intensity, spectral distribution curve, and chromaticity characteristic transmission information such as transmission correlation parameters with other regions.

[0025] Step S122: Extract the initial conduction characteristics of the body region of each light-emitting unit. The initial conduction characteristics reflect the conduction initiation attribute, conduction intensity benchmark, and initial conduction direction preference of the chromaticity characteristics of the light-emitting unit.

[0026] After region segmentation, initial conduction features are extracted for the body region of each emitting unit. First, statistical analysis is performed on the spectral data of all pixels within the body region to calculate basic spectral parameters such as average spectral intensity, peak wavelength, and full width at half maximum (FWHM). These parameters form the basis of the conduction initiation attributes. The conduction intensity benchmark is determined by calculating the integral value of the spectral intensity within the body region, covering the entire visible spectrum. This integral value reflects the initial energy level of the emitting unit's chromaticity features conducted to its surroundings. Determining the initial conduction direction preference requires analyzing the spatial distribution gradient of spectral intensity within the body region. By calculating the rate of change of spectral intensity in different directions, the direction with the largest rate of change is identified; this direction is the initial conduction direction preference. For example, if the rate of change of spectral intensity on the right side of a emitting unit's body region is greater than in other directions, its initial conduction direction preference is to the right. Integrating the above conduction initiation attributes, conduction intensity benchmark, and initial conduction direction preference forms the initial conduction features for each emitting unit's body region. These initial conduction features are stored in vector form, containing parameters of multiple dimensions, and reflect the initial state of the emitting unit's chromaticity feature conduction.

[0027] Step S1221: Perform pixel-level fine scanning on the body area of ​​each light-emitting unit to collect the chromaticity values, brightness distribution information and chromaticity correlation data of all pixels in the body area.

[0028] In this embodiment, a high-resolution image scanning device is used to perform pixel-level fine scanning of the body area of ​​each light-emitting unit. The scanning accuracy of the device reaches 0.1μm, ensuring that detailed information of each pixel can be captured. During the scanning process, for each pixel within the body area, its chromaticity values ​​in the three primary color channels (red, green, and blue) and its luminance value are collected. Simultaneously, the chromaticity correlation data between pixels is determined by calculating the differences in chromaticity values ​​and luminance between adjacent pixels. For example, for two adjacent pixels, the Euclidean distance between their red, green, and blue chromaticity values ​​is calculated; the smaller the distance, the higher the chromaticity correlation between the two. The collected chromaticity values, luminance distribution information, and chromaticity correlation data of all pixels are stored in a two-dimensional array. The rows and columns of the array correspond to the horizontal and vertical coordinates of the pixel within the body area, respectively, and the array elements contain various data for that pixel.

[0029] Step S1222: Using the geometric center of the light-emitting unit body area as a reference point, divide it into multiple concentric annular regions. Each annular region contains pixels at the same radial distance from the reference point.

[0030] After acquiring the coordinate data of all pixels within the body region, the geometric center of the body region is used as the reference point. First, the coordinates of the geometric center of the body region are determined by calculating the coordinates of the boundary pixels. Then, based on the size and pixel density of the body region, multiple different radial distance values ​​are set, and multiple concentric annular regions are divided according to these radial distance values, centered on the reference point. For example, for a square body region with a side length of 100 pixels and a geometric center coordinate of (50, 50), radial distances of 10 pixels, 20 pixels, and 30 pixels can be set to divide it into three concentric annular regions. Each annular region has a width of 10 pixels and contains pixels 10, 20, and 30 pixels away from the reference point, respectively. Pixels within each annular region have the same radial distance, facilitating subsequent analysis of the distribution of chromaticity features at different radial positions.

[0031] Step S1223: Analyze the concentration of chromaticity value distribution of pixels within each annular region, and extract the dominant chromaticity attribute of each annular region. The dominant chromaticity attribute is the most representative chromaticity feature within the annular region.

[0032] For each concentric annular region, the chromaticity value distribution of all pixels within it is statistically analyzed. A clustering algorithm is used to group pixels with similar chromaticity values ​​into one class, and then the proportion of pixels in each class is calculated. The chromaticity value corresponding to the class with the highest proportion is the dominant chromaticity attribute of that annular region. For example, if most pixels in a certain annular region have high red chromaticity values ​​and low green and blue chromaticity values, then the dominant chromaticity attribute of that annular region is red. By extracting the dominant chromaticity attribute of each annular region in this way, the main chromaticity characteristics at different radial positions can be reflected.

[0033] Step S1224: Calculate the gradient of the dominant chromaticity attribute change from the central annular region to the edge annular region, and determine the distribution characteristics and diffusion initiation characteristics of chromaticity features within the body region.

[0034] After obtaining the dominant chromaticity attribute of each annular region, the difference in dominant chromaticity attribute between adjacent annular regions is calculated. The ratio of this difference to the radial distance between the annular regions is the gradient of the dominant chromaticity attribute change. Starting from the central annular region, the gradients are calculated sequentially towards the edge annular regions, forming a gradient sequence. By analyzing this gradient sequence, the distribution characteristics of chromaticity features within the body region can be determined. For example, a large gradient value indicates that the chromaticity feature changes drastically in the radial direction and is unevenly distributed; a small gradient value indicates that the chromaticity feature is relatively evenly distributed. Simultaneously, the direction of the gradient also reflects the diffusion initiation characteristics of the chromaticity feature: a positive gradient indicates that the dominant chromaticity attribute is changing in an increasing direction, while a negative gradient indicates a decreasing direction.

[0035] Step S1225: Extract the core chromaticity parameters of the central annular region. These core chromaticity parameters are the essential manifestation of the chromaticity characteristics of the luminescent unit and serve as the basis for calculating the conduction intensity benchmark.

[0036] The central annular region is the core part of the light-emitting unit's body area, and its chromaticity characteristics best represent the essential attributes of the light-emitting unit. Therefore, the average chromaticity values ​​of all pixels within the central annular region are extracted as the core chromaticity parameter. Specifically, the red, green, and blue chromaticity values ​​of each pixel within the central annular region are arithmetically averaged to obtain three average chromaticity values. These three values ​​together constitute the core chromaticity parameter. This core chromaticity parameter reflects the most fundamental chromaticity characteristics of the light-emitting unit.

[0037] Step S1226: Based on the core chromaticity parameters and the dominant chromaticity attributes of the edge annular region, calculate the initial transmission intensity benchmark value, which reflects the initial intensity level of the chromaticity features transmitted outward.

[0038] First, the core chromaticity parameters and the dominant chromaticity attributes of the edge annular region are converted into a unified chromaticity space, such as the CIEXYZ chromaticity space. Then, the Euclidean distance between the core chromaticity parameters and the dominant chromaticity attributes of the edge annular region is calculated; this distance reflects the degree of difference between the two. This distance value is normalized to a range between 0 and 1. Finally, the normalized distance value is multiplied by a preset intensity coefficient to obtain the initial conduction intensity reference value. The magnitude of the intensity coefficient is set according to the characteristics of the minimized display and the actual application requirements. For example, for high-brightness displays, the intensity coefficient can be appropriately increased to improve the initial conduction intensity reference value.

[0039] Step S1227: Analyze the spatial gradient distribution of pixel chromaticity values ​​within the body region to determine the diffusion potential energy difference of chromaticity features in different directions, wherein the diffusion potential energy difference reflects the initial conduction direction preference.

[0040] The chromaticity gradient of each pixel within the body region is calculated in four directions: horizontal, vertical, 45 degrees, and 135 degrees. For each direction, the gradient value is obtained by calculating the ratio of the chromaticity difference between adjacent pixels to their distance. Then, the average gradient value of the entire body region in each direction is calculated. The larger the average gradient value, the more drastic the change in chromaticity characteristics in that direction, and the greater the diffusion potential. By comparing the average gradient values ​​in the four directions, the direction with the greatest diffusion potential is determined, and this direction is the initial propagation direction preference. For example, if the average gradient value is largest in the horizontal direction, then the initial propagation direction preference is the horizontal direction.

[0041] Step S1228: Perform an integration operation on the core chromaticity parameters, the initial conduction intensity reference value, the initial conduction direction preference, and the gradient of the dominant chromaticity attribute change in the annular region to form the basic components of the initial conduction characteristics.

[0042] The core chromaticity parameters (containing red, green, and blue components), the initial conduction intensity baseline value, the initial conduction direction preference (represented by a direction vector), and the gradient of the dominant chromaticity attribute change in the annular regions (containing gradient values ​​of multiple annular regions) are combined in a specific order to form a multidimensional vector. For example, the first three elements of the vector are the red, green, and blue components of the core chromaticity parameters; the fourth element is the initial conduction intensity baseline value; the fifth to eighth elements are the direction vector components of the initial conduction direction preference; and the subsequent elements are the gradient values ​​of the dominant chromaticity attribute change in each annular region. This multidimensional vector is the basic component of the initial conduction characteristics, containing key information reflecting the initial state of chromaticity characteristic conduction of the luminescent unit.

[0043] Step S1229: Remove redundant information from the basic components that is irrelevant to cross-unit transmission, and retain the key elements that can accurately reflect the transmission initiation attributes.

[0044] Correlation analysis was used to calculate the correlation coefficient between each component of the basic elements and the cross-unit transmission effect. The Pearson correlation coefficient method was employed; for each component, it was compared with the actual measured value of the cross-unit transmission effect to obtain the correlation coefficient. A correlation coefficient threshold was set, and components with correlation coefficients below this threshold were considered redundant information unrelated to cross-unit transmission and were discarded. For example, if the correlation coefficient between the dominant chromaticity attribute change gradient of a certain annular region and the cross-unit transmission effect was low, this gradient value was removed from the basic elements. The retained components are the key elements that accurately reflect the initiation attribute of transmission.

[0045] Step S1210: Organize the key elements in a structured manner according to the logical order of core chromaticity parameters, conduction intensity benchmark, conduction direction preference, and regional change gradient to form initial conduction features. Normalize and perform principal component analysis on the initial conduction feature data after structured organization. Retain principal components whose variance contribution rate exceeds the preset threshold and perform standardization scaling. Perform feature recognition enhancement operation and stability adaptation operation.

[0046] After removing redundant information, the key elements are arranged in the order of core chromaticity parameters, conduction intensity benchmark, conduction direction preference, and regional variation gradient to form a structured data structure, namely the initial conduction features. Then, this initial conduction feature data is normalized, converting the value range of each element to between 0 and 1 to eliminate dimensional differences between different elements. The normalization process uses a min-max normalization method, where the value of each element is subtracted from its minimum value, and then divided by the difference between its maximum and minimum values. After normalization, principal component analysis is performed on the data to calculate the variance contribution rate of each principal component. Principal components with a variance contribution rate exceeding a preset threshold (e.g., 85%) are retained to reduce data dimensionality and retain key information. Next, the retained principal components are standardized and scaled to have a mean of 0 and a standard deviation of 1, further improving data stability and comparability. Finally, feature recognition enhancement operations are performed by increasing the weight of key features, and stability adaptation operations are performed through smoothing to ensure that the initial conduction features accurately and stably reflect the conduction initiation attributes of the luminescent unit.

[0047] Step S123: Identify the directly adjacent and non-directly adjacent light-emitting units of each light-emitting unit, and record the hierarchical relationship and physical location distribution of the directly transmitted and indirectly transmitted objects.

[0048] Based on the physical layout information of the Miniled display, each light-emitting unit has a fixed coordinate position. Directly adjacent and indirectly adjacent relationships are determined by comparing the coordinates of each light-emitting unit with the coordinates of other light-emitting units. Directly adjacent light-emitting units are those whose coordinates differ by one light-emitting unit spacing in the horizontal or vertical direction. For example, for a light-emitting unit with coordinates (x, y), its directly adjacent light-emitting units have coordinates of (x+1, y), (x-1, y), (x, y+1), and (x, y-1) (there may be fewer than four at the display boundary). Indirectly adjacent light-emitting units are those whose coordinates differ by one light-emitting unit spacing in the diagonal direction, i.e., units with coordinates of (x+1, y+1), (x+1, y-1), (x-1, y+1), and (x-1, y-1). After identifying these adjacent units, the hierarchical relationship between directly transmitted objects (directly adjacent light-emitting units) and indirectly transmitted objects (indirectly adjacent light-emitting units) is recorded. Directly transmitted objects are at the first level, and indirectly transmitted objects are at the second level. At the same time, their physical location distribution, i.e., the coordinate information of each adjacent unit, is recorded so that the transmission trajectory can be tracked and the transmission link can be constructed later.

[0049] Step S124: Track the forward conduction trajectory of the initial conduction characteristics from the light-emitting unit body region through the gap region to the directly adjacent light-emitting unit, and determine the direction, length and conduction resistance distribution of the direct conduction path.

[0050] Chromaticity feature data from the inter-cell transmission image recorded in the inter-cell transmission region is used to trace the transmission trajectory of the initial transmission feature. Starting from the edge of the luminescent cell body region, a series of feature points are identified within the inter-cell region along the direction of gradual diffusion of the chromaticity feature; the line connecting these feature points forms the forward transmission trajectory. The direction of the direct transmission path is determined based on the coordinate changes of these feature points. For example, starting from the right edge of the body region, if the feature points move to the right, the path is horizontal to the right. The length of the transmission path is determined by calculating the straight-line distance between the starting and ending feature points, in micrometers. Determining the transmission resistance distribution requires considering the physical properties of the inter-cell region, such as the refractive index and thickness of the medium. By analyzing the attenuation of chromaticity feature intensity at different locations within the inter-cell region, the transmission resistance value at each location is calculated, forming a transmission resistance distribution curve. For example, at the center of the inter-cell region, the transmission resistance is low due to the uniformity of the medium, while it is high near the edge.

[0051] Step S125: Analyze the indirect conduction trajectory of the initial conduction feature from the directly adjacent light-emitting unit to the non-directly adjacent light-emitting unit, and determine the jump node, conduction delay and feature attenuation characteristics of the indirect conduction path.

[0052] When the initial conduction feature is transmitted to a directly adjacent emitting unit, it diffuses and attenuates within that unit before being transmitted to a non-directly adjacent emitting unit. This process is called secondary conduction. The indirect conduction trajectory is traced by analyzing the changes in chromaticity features between directly adjacent and non-directly adjacent emitting units in the cross-unit conduction image of chromaticity features. A jump node is the starting point of the initial conduction feature's transmission from a directly adjacent emitting unit to a non-directly adjacent emitting unit, typically located at the edge region of the directly adjacent emitting unit. The conduction delay is the time required for the initial conduction feature to travel from the body region of the emitting unit, through the directly adjacent emitting unit, and then to the non-directly adjacent emitting unit, calculated by recording the timestamps of feature transmission. Feature attenuation is determined by comparing the intensity values ​​of the initial conduction feature before and after the jump node, calculating the percentage of intensity attenuation, and analyzing the variation of attenuation with the conduction distance.

[0053] Step S1251: Collect real-time chromaticity state change data of directly adjacent light-emitting units after receiving the initial conduction feature, and determine the conduction diffusion range and stabilization time of the initial conduction feature in directly adjacent units.

[0054] Real-time monitoring equipment is used to continuously monitor the chromaticity state of directly adjacent light-emitting units after receiving the initial conductive feature. The monitoring equipment collects chromaticity values ​​and luminance distribution data of the unit at millisecond intervals, forming time-series data. By analyzing this time-series data, the diffusion range of the initial conductive feature within directly adjacent units is determined, i.e., the maximum area that the chromaticity feature can reach from the receiving point outwards. The stabilization time refers to the time required from the initial conductive feature being received until the chromaticity state of the unit no longer changes significantly; it is determined by judging whether the chromaticity value changes at multiple consecutive time points are less than a preset threshold.

[0055] Step S1252: Starting from the stable region of conduction and diffusion within directly adjacent light-emitting units, trace the transmission trajectory of chromaticity features to non-directly adjacent light-emitting units, and mark the coordinates of all key points on the trajectory and the corresponding chromaticity feature data.

[0056] After the conduction and diffusion of directly adjacent light-emitting units reach stability, the boundary of the stable region is used as the starting point to continue tracing the transmission trajectory of chromaticity features. By analyzing the changes in chromaticity data between this region and non-directly adjacent light-emitting units in the cross-unit transmission image of chromaticity features, key locations on the transmission trajectory are determined. These key locations include inflection points on the trajectory and points where the intensity of chromaticity features changes significantly. For each key location, its coordinates on the display screen and the chromaticity feature data at that point, such as the intensity values ​​of the red, green, and blue components, are recorded.

[0057] Step S1253: Based on the marked key location points, identify the jump positions in the process of chromaticity features being transferred from directly adjacent units to non-directly adjacent units. These jump positions serve as jump nodes in the indirect transmission path, and the spatial coordinates and feature transformation status of the jump nodes are recorded.

[0058] Analyze the marked key points to identify locations where the direction of chromaticity feature transmission abruptly changes. These locations are the jump points, or jump nodes in the indirect transmission path. For example, if the chromaticity feature was originally transmitted horizontally but suddenly turns diagonally at a key point, that point is the jump node. Record the spatial coordinates of the jump nodes and analyze the changes in chromaticity feature data before and after the jump to determine the feature transformation state, such as increases or decreases in feature intensity and changes in spectral distribution.

[0059] Step S1254: Calculate the total time required for the initial conduction characteristics to jump from the light-emitting unit body region through directly adjacent units to non-directly adjacent units, and subtract the time for direct conduction to adjacent units to obtain the conduction delay time for indirect conduction.

[0060] The total time and direct conduction time are calculated by recording the time it takes for the initial conduction feature to travel from the luminescent unit's body region, to reach the directly adjacent unit, and to reach the non-directly adjacent unit. The total time refers to the time interval from departure to arrival at the non-directly adjacent unit, while the direct conduction time refers to the time interval from departure to arrival at the directly adjacent unit. The difference between the two is the conduction delay of indirect conduction, which reflects the additional time spent in the secondary conduction process.

[0061] Step S1255: Analyze the change range of chromaticity feature data before and after the jump node, and determine the conversion loss and attribute change characteristics of the features during the jump process.

[0062] Compare the chromaticity feature data of key points before and after the jump node, and calculate the intensity change of each component (red, green, and blue). The change is calculated as a percentage: (intensity after jump - intensity before jump) / intensity before jump × 100%. A negative change indicates a decrease in feature intensity, i.e., conversion loss exists; a positive change indicates an increase in feature intensity. Simultaneously, analyze the changes in the spectral distribution curves before and after the jump to determine attribute change characteristics, such as the shift in peak wavelength and changes in half-width at half-maximum (FWHM).

[0063] Step S1256: Divide the indirect transmission trajectory into multiple detection segments at equal distances, and extract the chromaticity feature intensity data of the start and end points of each detection segment.

[0064] Based on the total length of the indirect transmission trajectory, it is divided into multiple detection segments of equal length. For example, if the total trajectory length is 100 micrometers, it can be divided into 10 detection segments, each 10 micrometers long. For each detection segment, the chromaticity characteristic intensity data of its start and end points are extracted, including the intensity values ​​of the red, green, and blue components.

[0065] Step S1257: Calculate the attenuation of the chromaticity characteristic intensity in each detection segment, and determine the local attenuation coefficient of the detection segment by combining the length of the detection segment and the characteristics of the conductive medium.

[0066] For each detection segment, the difference between the chromaticity characteristic intensity at the endpoint and the starting point is calculated to obtain the intensity attenuation within that segment. Then, considering the length of the detection segment and the characteristics of the conducting medium (such as the absorption coefficient, scattering coefficient, etc.), the local attenuation coefficient is determined as follows: Local attenuation coefficient = Attenuation / (Detection segment length × Medium characteristic parameter). The medium characteristic parameter is set according to the actual medium type in the gap region. For example, the characteristic parameter for air is 1, while the characteristic parameter for a specific material medium may be greater than 1.

[0067] Step S1258: Integrate the local attenuation coefficients of all detection segments, analyze their variation characteristics with conduction distance and jump node distribution, and summarize the overall characteristic attenuation characteristics of the indirect conduction path.

[0068] The local attenuation coefficients of all detection segments are arranged in order of conduction distance, and curves showing the change of attenuation coefficients with conduction distance are plotted. Simultaneously, the positions of jump nodes on the curves are marked, and the changes in attenuation coefficients before and after the jump nodes are analyzed. By analyzing the shape and trend of the curves, the overall characteristic attenuation features of the indirect conduction path are summarized, such as whether the attenuation coefficient increases linearly with increasing conduction distance, and whether jump nodes cause sudden changes in the attenuation coefficient.

[0069] Step S1259: Use a piecewise function to fit the overall feature decay feature, generate a mathematical model of the decay feature, and perform a precise description and adaptation operation of the decay feature.

[0070] Based on the overall characteristic attenuation trend, the model is divided into multiple linear segments, each corresponding to a conduction region between jump nodes. For each linear segment, a linear function is used for fitting, yielding the relationship between the attenuation coefficient and the conduction distance. These piecewise functions are combined to form a mathematical model of the overall characteristic attenuation. Then, by adjusting the function parameters, the model accurately fits the actual attenuation characteristic data, performing a precise attenuation characteristic description adaptation operation to ensure that the model truly reflects the characteristic attenuation of the indirect conduction path.

[0071] Step S1260: Link and store the jump node coordinates, conduction delay duration, feature transformation state, and fitted overall attenuation features to form complete descriptive data of the indirect conduction trajectory.

[0072] The determined jump node coordinates, conduction delay duration, feature transition states, and the fitted overall attenuation feature mathematical model are organized and stored according to a specific format. For example, a structure data type can be used, containing fields such as an array of jump node coordinates, conduction delay duration values, a string describing the feature transition states, and parameters of the attenuation feature mathematical model. By storing these data in association, a complete description of the indirect conduction trajectory is formed.

[0073] Step S1261: Based on the complete description data, construct an indirect transmission path model, which simulates the entire process of chromaticity features being transmitted to non-directly adjacent units via secondary transmission.

[0074] An indirect conduction path model is constructed using the coordinates of the jump nodes, conduction delay duration, feature transition states, and overall attenuation features from the complete descriptive data. This model takes the intensity and direction of the initial conducted features as input and simulates the transmission process of features between jump nodes, including conduction delay, feature attenuation, and transitions. The final output is the intensity and state of the chromaticity features reaching non-directly adjacent emitting units. The model is constructed using a physics-based modeling approach, combining mathematical models and empirical parameters to ensure accurate simulation of the actual indirect conduction process.

[0075] Step S1262: Compare the predicted trajectory of the indirect conduction path model with the standard trajectory of different conduction scenarios. If the similarity is lower than a preset threshold, adjust the model parameters and perform the adaptation operation between the indirect conduction path model and the trajectory simulation standard of different conduction scenarios.

[0076] Standard trajectory data for different conduction scenarios are collected. These standard trajectories are ideal conduction trajectories obtained through actual measurements. The trajectories predicted by the indirect conduction path model are compared with these standard trajectories, and the similarity between them is calculated. The similarity calculation uses a dynamic time warping algorithm, which evaluates the similarity by aligning the two trajectories and calculating the sum of the distances between corresponding points. If the similarity is lower than a preset threshold, it indicates that the model's prediction results deviate significantly from the actual situation. The model's parameters, such as the attenuation coefficient and conduction delay time, need to be adjusted, and the simulation and comparison should be repeated until the similarity between the model's predicted trajectory and the standard trajectory reaches or exceeds the preset threshold. At this point, the model is adapted to the trajectory simulation standards for different conduction scenarios.

[0077] Step S126: Calculate the link impedance of each direct and indirect conduction path, and assign a conduction efficiency coefficient to each path based on the link impedance to form a set of conduction paths after impedance matching.

[0078] Link impedance calculation considers factors such as the length of the conduction path, the resistance and capacitance of the conducting medium, etc. For direct conduction paths, the link impedance is mainly determined by the dielectric resistance and capacitance of the gap region, which can be calculated by measuring the voltage and current at both ends of the path. For indirect conduction paths, in addition to the dielectric resistance and capacitance, the contact resistance and reflection loss at the jump nodes also need to be considered. The total link impedance is obtained by adding the impedances of each part. The conduction efficiency coefficient is inversely proportional to the link impedance; that is, the smaller the link impedance, the larger the conduction efficiency coefficient. The specific calculation formula is: Conduction efficiency coefficient = Reference impedance / Link impedance, where the reference impedance is a preset reference impedance value. The calculated conduction efficiency coefficient is assigned to the corresponding conduction path to form an impedance-matched set of conduction paths. Each path in this set of conduction paths has a corresponding conduction efficiency coefficient, which is used to subsequently construct the forward conduction link network.

[0079] Step S127: Based on the set of conduction paths after impedance matching and the jump nodes, build a forward conduction link network covering the cross-unit conduction paths of all light-emitting units.

[0080] Using each emitting unit as a node, nodes with both direct and indirect conduction paths are connected based on the set of conduction paths after impedance matching. During the connection process, jump nodes act as intermediate nodes in indirect conduction paths, connecting different direct conduction paths to form a complex network structure. The forward conduction link network is constructed using graph theory, representing emitting units as vertices of the graph, conduction paths as directed edges, and edge weights as conduction efficiency coefficients. Through this method, a directed graph covering all emitting units is constructed, i.e., the forward conduction link network. This network can demonstrate the forward conduction paths and efficiencies of chromaticity features across the entire display screen.

[0081] Step S128: Extract the conduction characteristic feedback information received by each light-emitting unit from other units, and determine the start point, end point and feedback transmission efficiency of the reverse feedback path.

[0082] When a light-emitting unit receives a conducted feature from another unit, it generates feedback information based on its own chromaticity state. This feedback information is extracted by monitoring changes in the output signal of each light-emitting unit. The reverse feedback path starts at the light-emitting unit receiving the conducted feature and ends at the original light-emitting unit that sent the feature. The calculation of feedback transmission efficiency is similar to that of forward transmission efficiency, considering the link impedance of the feedback path and the attenuation of the feedback signal. It is determined by measuring the ratio of the strength of the feedback signal to the strength of the original conducted feature. For example, if the strength of the original conducted feature is A and the strength of the feedback signal is B, then the feedback transmission efficiency is B / A.

[0083] Step S129: Map the start and end points of the reverse feedback path to the corresponding nodes in the forward propagation link network to construct a bidirectional link framework that includes forward propagation and reverse feedback.

[0084] Based on the forward propagation link network, the start and end points of each reverse feedback path are mapped to the corresponding emitting unit nodes in the network. For each reverse feedback path, a directed edge from the start point to the end point is added to the forward propagation link network, with the edge weight being the feedback transmission efficiency. In this way, the directed edges of forward propagation and the directed edges of reverse feedback together constitute a bidirectional link framework that includes both forward propagation and reverse feedback, enabling the bidirectional transmission of chromaticity features between emitting units.

[0085] Step S1210: Deploy a conduction feature detection node in the bidirectional link framework. The conduction feature detection node captures the intersection state and superposition effect of positive conduction features and negative feedback features in real time.

[0086] Conductive feature detection nodes are deployed at key locations within the bidirectional link framework. These locations include intersections of forward and reverse feedback paths, hop nodes, and intermediate nodes of the link. Each detection node integrates miniature sensors and a data processing unit, enabling real-time acquisition of information such as the intensity, phase, and spectrum of the forward and reverse feedback features flowing through it. Analysis of this information captures the intersection status of the two features, such as whether they arrive simultaneously and whether their intensities overlap, and calculates the overlap effect, such as the resulting intensity and new spectral distribution. The detection nodes transmit the captured data to the central processing unit in real-time for subsequent link parameter updates.

[0087] Step S1211: Based on the capture results of the detection nodes, update the transmission priority parameters and feedback response rate parameters of the link to form a closed-loop link for cross-unit transmission of chromaticity features.

[0088] After receiving data from the detection nodes, the central processing unit analyzes the intersection and superposition effects of forward propagation features and reverse feedback features. If it finds that the superposition of features on a certain forward propagation path with the reverse feedback features results in excessive intensity, potentially leading to chromaticity unevenness, the propagation priority parameter of that path is reduced, decreasing its weight in the propagation process. Simultaneously, based on the transmission time and intensity changes of the feedback features, the feedback response rate parameter is adjusted to accelerate or slow down the transmission speed of the feedback signal, ensuring the dynamic balance of the link. By continuously updating the propagation priority parameter and the feedback response rate parameter, the bidirectional link framework can adaptively adjust the propagation and feedback processes, ultimately forming a stable closed-loop link for cross-unit chromaticity feature propagation.

[0089] Step S130: Based on the chromaticity feature transmission characteristics under different working scenarios, define the chromaticity feature transmission threshold, transmission attenuation coefficient and cross-over superposition adaptation method for each light-emitting unit, and generate a dynamic adaptation transmission rule set.

[0090] Different working scenarios, such as varying ambient light intensities, the type of image content displayed on the screen (e.g., static images, dynamic videos), and the screen's operating temperature, will have different impacts on the transmission of chromaticity features. In this embodiment, several typical working scenarios are first preset, such as displaying dynamic videos in a bright environment and displaying static images in a dark environment. For each scenario, chromaticity feature transmission images are collected experimentally to analyze the variation patterns of the transmission features. Then, based on these patterns, a transmission threshold is defined for each light-emitting unit; that is, cross-unit transmission is only allowed when the intensity of the transmission feature exceeds this threshold. The transmission attenuation coefficient is set according to the attenuation of the transmission features in different scenarios to correct for intensity loss during transmission. The cross-over superposition adaptation method specifies the processing method when multiple transmission features are superimposed, such as linear superposition or taking the maximum value. By associating the above parameters and methods with the corresponding working scenarios, a dynamic adaptation transmission rule set is formed. When the working scenario changes, the system can automatically call the corresponding rules for transmission control.

[0091] Step S131: Acquire standard chromaticity feature transmission images of the Miniled display screen under various preset working scenarios. The standard chromaticity feature transmission images record the cross-unit transmission characteristics of the ideal chromaticity uniformity state under different scenarios.

[0092] The preset working scenarios include bright environments with ambient light intensities of 100 lux, 500 lux, and 1000 lux, a dark environment with 0 lux, and scenarios displaying different content types such as dynamic videos with rich colors, grayscale images, and single-color images. In each scenario, the minimized display screen is adjusted to an ideal chromaticity uniformity state, that is, the chromaticity output of all light-emitting units on the display screen is kept consistent through professional calibration equipment. Then, the same spectral imaging equipment as in step S110 is used to acquire the chromaticity feature cross-unit transmission image at this time, which serves as the standard chromaticity feature transmission image. The aforementioned standard image records the cross-unit transmission characteristics of chromaticity features in different scenarios under the ideal uniformity state, including transmission paths, intensity distribution, and superposition effects.

[0093] Step S132: For the standard chromaticity feature transmission image in each scene, apply image segmentation and feature recognition algorithms to obtain the standard initial transmission features, standard transmission paths and standard superposition effects of all light-emitting units in that scene.

[0094] For the standard chromaticity feature transmission image of each scene, the same U-Net-based image segmentation model as in step S121 is first used for region segmentation to separate the body region, gap region, and indirect transmission influence region of each emitting unit. Then, for the body region, standard initial transmission features are extracted according to the method in step S122, including core chromaticity parameters, transmission intensity benchmark, and transmission direction preference. For the transmission path, standard transmission paths, including direct and indirect transmission paths, are identified by analyzing the chromaticity feature data of the gap region and the indirect transmission influence region, and their direction, length, link impedance, and other parameters are recorded. The standard superposition effect is determined by analyzing the chromaticity data after multiple transmission features are superimposed in the same region, such as the superimposed intensity and spectral distribution. The above standard initial transmission features, standard transmission paths, and standard superposition effects are associated and stored with the corresponding scene.

[0095] Step S133: Analyze the variation characteristics of the standard initial conduction characteristics on different conduction paths under each scenario, and define the standard conduction threshold of each light-emitting unit under the scenario. The standard conduction threshold is the critical value of conduction intensity corresponding to chromaticity uniformity.

[0096] For each scenario, the intensity changes of the initial standard conduction feature are tracked along each standard conduction path. Starting from the body region of the emitting unit, the intensity of the conduction feature is recorded at different locations along the conduction path. When the intensity of the conduction feature decreases to a certain value, continued conduction will no longer effectively affect the chromaticity of other emitting units; this value is the standard conduction threshold. The determination of the standard conduction threshold needs to consider the requirement of maintaining chromaticity uniformity, ensuring that conduction features above this threshold can be effectively utilized, while conduction features below this threshold are ignored to reduce unnecessary conduction interference. The standard conduction threshold for each emitting unit may differ in different scenarios and needs to be defined separately according to the characteristics of each scenario.

[0097] Step S134: Calculate the standard attenuation magnitude of the standard conduction feature on the direct and indirect conduction paths for each scenario, and define the standard conduction attenuation coefficient corresponding to different paths based on the attenuation magnitude.

[0098] In each scenario, for each direct and indirect conduction path, the intensity attenuation of the standard conduction feature from the starting point to the ending point is calculated. The attenuation is calculated by subtracting the ending point intensity from the starting point intensity. Then, the attenuation is divided by the length of the conduction path to obtain the attenuation per unit length. The standard conduction attenuation coefficient is defined as the ratio of the attenuation per unit length to the starting point intensity, i.e., standard conduction attenuation coefficient = (starting point intensity - ending point intensity) / (starting point intensity × path length). This standard conduction attenuation coefficient reflects the attenuation ratio of the conduction feature per unit length of path. For direct and indirect conduction paths, due to different path characteristics, their standard conduction attenuation coefficients are also different and need to be calculated and defined separately.

[0099] Step S135: Analyze the standard pattern of cross-over superposition of multiple light-emitting unit conduction features in each scenario, and define the feature complementarity rules and conflict resolution methods in the superposition process.

[0100] In regions where the conduction characteristics of multiple luminescent units overlap, the chromaticity data of these regions in the standard chromaticity feature conduction image are analyzed to summarize the standard overlay patterns. Feature complementarity rules refer to how to overlay different conduction characteristics when their spectral components are complementary to enhance the overall chromaticity effect; for example, red and green features overlap to form yellow features. Conflict resolution methods refer to how to handle conflicts in the intensity or spectrum of different conduction characteristics; for example, when the overlay of two red features with similar intensities results in excessive intensity, averaging is used to resolve the conflict. By analyzing the standard patterns, the complementarity rules and conflict resolution methods to be followed under different overlay conditions are clarified to ensure that the overlaid chromaticity features meet the uniformity requirements.

[0101] Step S136: Assign a higher weight coefficient to the scene parameters of ambient lighting conditions, display workload, and target display content type than other scene parameters. Based on the scene parameters after weight assignment, establish a correspondence model between the scene parameters and the standard conduction threshold, standard conduction attenuation coefficient, and standard superposition method. The correspondence model reflects the influence of different scene parameters on the conduction rules.

[0102] Among numerous scene parameters, ambient lighting conditions, display workload (such as the number of light-emitting units per unit area and working time), and the type of target display content have the most significant impact on chromaticity feature transmission. Therefore, these parameters are assigned relatively high weighting coefficients. For example, the weighting coefficient for ambient lighting conditions is 0.4, for display workload it is 0.3, for target display content type it is 0.2, and the sum of other parameters is 0.1. Then, using the scene parameters with the above weighting as input, and the standard transmission threshold, standard transmission attenuation coefficient, and standard superposition method as output, a corresponding relationship model is established.

[0103] Step S1361: Classify and organize the collected scene parameters under various preset working scenarios, determine the type, scope of influence and quantification method of each scene parameter, and form a scene parameter set.

[0104] The collected scene parameters are categorized according to ambient lighting conditions, display workload, and target content type. Ambient lighting conditions are continuous parameters, affecting the color perception of the entire display screen, and are quantified as illuminance (lux). Display workload is a discrete parameter, affecting the heating and aging of the light-emitting units, and is quantified as average power consumption (W) per unit time. Target content type is a categorical parameter, affecting the color characteristic distribution of different areas, and is quantified as content category encoding (e.g., 0 for static images, 1 for dynamic videos, etc.). Each parameter is described and quantified in detail to form a scene parameter set.

[0105] Step S1362: Extract the standard conduction threshold, standard conduction attenuation coefficient and standard superposition method of all light-emitting units in each scenario, and construct a standard rule dataset, wherein the standard rule dataset corresponds one-to-one with the scenario parameter set.

[0106] From the processing results of steps S133, S134, and S135, extract the standard conduction threshold, standard conduction attenuation coefficient, and standard superposition method for all emitting units in each scenario. Arrange the above data according to the coordinate order of the emitting units to form a regular data matrix, where each row corresponds to the regular parameters of one emitting unit. Combine the regular data matrices of all scenarios to construct a standard regular dataset. This standard regular dataset corresponds one-to-one with each scenario in the scenario parameter set, that is, each scenario parameter vector corresponds to one regular data matrix.

[0107] Step S1363: Use a feature mapping algorithm to convert each scene parameter in the scene parameter set into a high-dimensional feature vector. This high-dimensional feature vector represents the essential attributes and influence weights of the scene parameter.

[0108] For continuous parameters (such as ambient light intensity), normalization is applied followed by equal-interval sampling, mapping them to multiple dimensions. For discrete parameters (such as display workload), one-hot encoding is used to convert them into binary vectors. Similarly, for categorical parameters (such as target display content type), one-hot encoding is also used to convert them into binary vectors. Then, based on the weight coefficients of each parameter, the converted vectors are weighted and combined to form a high-dimensional feature vector. For example, ambient light intensity, after mapping, becomes a 10-dimensional vector, display workload becomes a 5-dimensional vector, and target display content type becomes a 3-dimensional vector. This weighted combination forms an 18-dimensional high-dimensional feature vector, which contains the essential attributes and influencing weight information of the scene parameters.

[0109] Step S1364: Normalize the standard conduction threshold and standard conduction attenuation coefficient in the standard rule dataset, and convert the standard superposition method into a standardized encoding vector to ensure that the rule data has a unified calculation dimension.

[0110] For the standard conduction threshold and standard conduction attenuation coefficient, a min-max normalization method is used to convert their numerical range to between 0 and 1, making them comparable. For the standard superposition method, a unique code is assigned to each rule according to different superposition rules (such as linear superposition, taking the maximum value, taking the average value, etc.), for example, 001 represents linear superposition, 010 represents taking the maximum value, 011 represents taking the average value, etc., and these codes are converted into standardized binary code vectors. Through these processes, all data in the standard rule dataset have a uniform computational dimension and numerical range, facilitating subsequent model training.

[0111] Step S1365: Construct a multi-level neural network model, which includes an input layer, a hidden layer and an output layer. The input layer receives high-dimensional feature vectors of scene parameters, and the output layer outputs the corresponding standardized rule data.

[0112] This multi-layer neural network model employs a three-layer structure. The number of neurons in the input layer is the same as the dimension of the high-dimensional feature vector (e.g., 18 dimensions). The hidden layer consists of two layers: the first layer has 64 neurons, and the second layer has 32 neurons. The number of neurons in the output layer is determined by the dimension of the standardized rule data. For example, if the rule data for each light-emitting unit contains three parameters (threshold, attenuation coefficient, and superposition encoding), and the display screen has N light-emitting units, then the number of neurons in the output layer is 3×N. Fully connected layers are used between the input layer and the hidden layers, between the hidden layers, and between the hidden layers and the output layer.

[0113] Step S1366: Configure the hidden layers of the multi-level neural network model to use non-linear activation functions, set their hierarchical connection methods, and perform model feature extraction and mapping capability adaptation operations.

[0114] The first hidden layer uses the ReLU activation function to enhance the model's non-linear fitting ability; the second layer uses the LeakyReLU activation function to address the neuron death problem that may occur with the ReLU function. The hierarchical connection method is forward propagation, meaning information is passed sequentially from the input layer through the hidden layers to the output layer. Before model training, the connection weights are randomly initialized, and appropriate learning rates and batch sizes are set. Feature extraction and mapping capability adaptation operations are performed to ensure the model can effectively learn the complex mapping relationship between scene parameters and rule data.

[0115] Step S1367: Use the high-dimensional feature vector of scene parameters and the corresponding standardized rule data as training samples, input them into the multi-level neural network model for training, and minimize the deviation between the output of the multi-level neural network model and the standard rule data by adjusting the weight parameters and activation function of the multi-level neural network model.

[0116] The high-dimensional feature vectors derived from the transformation of the scene parameter set are used as input, and the corresponding standard rule dataset is used as output to form the training sample set. A gradient descent optimization algorithm is employed, with mean squared error as the loss function, to train a multi-level neural network model. During training, the connection weights and activation function parameters of the model are continuously adjusted using backpropagation to gradually reduce the deviation between the model's output rule data and the standard rule data. The number of training iterations is determined based on the model's convergence, continuing until the loss function value is less than a preset threshold.

[0117] Step S1368: During the training process, the cross-validation method is used to divide the training set and the validation set, and the prediction error of the model on the validation set is evaluated. The multi-level neural network model is used to collect prediction data and generalization data in different scenarios.

[0118] The training sample set was divided into a training set and a validation set in a 7:3 ratio. At the end of each training iteration, the model was evaluated using the validation set, and the mean absolute error and root mean square error (RMSE) between the predicted rule data and the standard rule data were calculated. Based on the evaluation results of the validation set, it was determined whether the model exhibited overfitting or underfitting, and the model complexity (e.g., increasing or decreasing the number of hidden layer neurons) or regularization parameters were adjusted accordingly. Simultaneously, prediction data and generalization data of the model in different scenarios were collected to analyze the model's adaptability to unseen scenarios.

[0119] Step S1369: Based on the feedback data of the validation set, adjust the hyperparameters of the multi-level neural network model, calibrate the model prediction output, and perform sensitivity adaptation operation of the multi-level neural network model to scene parameter changes and rule prediction data calibration operation.

[0120] Hyperparameters include learning rate, batch size, and the number of hidden layer neurons. Based on feedback data from the validation set, the optimal combination of hyperparameters is found using grid search or random search. For example, the learning rate can be selected from multiple values ​​such as 0.001, 0.01, and 0.1. By comparing the model performance under different hyperparameter combinations, the optimal hyperparameters are determined. Then, the model's predicted output is calibrated. For example, when there is a systematic deviation between the predicted standard transmission threshold and the actual value, a deviation compensation term is added for correction. Sensitivity adaptation to scene parameter changes is performed to ensure that the model's output rule data can smoothly transition when scene parameters change slightly; rule prediction data calibration is also performed to further improve the accuracy of model predictions.

[0121] Step S1370: After training is completed, input the test set into the multi-level neural network model, compare the error between the model output rule data and the actual standard rule data, and perform a consistency verification operation between the rule data output by the multi-level neural network model and the actual standard rule data.

[0122] The reserved test set is input into the trained multi-layer neural network model to obtain the predicted rule data. Various error metrics between the predicted data and the actual standard rule data are calculated, such as mean absolute error, root mean square error, and maximum error. If these error metrics are all within the preset acceptable range, the model's predictive performance is good, and it passes the consistency verification; otherwise, the model structure or training process needs to be re-examined, problems identified, and improvements made until the model passes the verification.

[0123] Step S1371: The multi-level neural network model that has passed the test is determined as the correspondence model between scene parameters and standard conduction rules. This correspondence model outputs the corresponding standard conduction threshold, standard conduction attenuation coefficient and predicted value of standard superposition method in real time according to the input scene parameters.

[0124] The multi-level neural network model that passed the testing and verification was determined as the final correspondence model. In practical applications, when the working scene of the Miniled display changes, the system collects the current scene parameters, converts them into high-dimensional feature vectors, and inputs them into the model. The model can output the predicted values ​​of the standard conduction threshold, standard conduction attenuation coefficient, and standard superposition method for each light-emitting unit in real time.

[0125] Step S1372: Compensate and correct the output of the corresponding relationship model based on historical error data to reduce the difference between the predicted value and the actual standard value.

[0126] Collect prediction error data from past model applications to analyze the distribution patterns and causes of these errors. For example, if the model's prediction error is large in a specific scenario, an error compensation model is established for that scenario. After the model outputs a predicted value, it is compensated and corrected based on historical error data; for example, the predicted value is increased by adding the corresponding error compensation value, thereby further reducing the difference between the predicted value and the actual standard value and improving the accuracy of the rule data.

[0127] Step S137: Collect scene parameter information in actual applications, including ambient lighting conditions, display workload, and key factors affecting color transmission such as the type of target display content.

[0128] During the actual operation of the minimized display, ambient lighting conditions are collected in real time by light sensors installed around the display to obtain light intensity values; workload information, such as current power consumption and running time, is obtained through the display's control system; and the displayed content is analyzed using image recognition algorithms to determine the type of target content, such as static images, dynamic videos, and text content. This information from the aforementioned key factors is collected in real time and transmitted to the central processing unit, serving as scene parameter input to the corresponding relationship model.

[0129] Step S138: Input the actual scene parameter information into the corresponding relationship model to generate the initial conduction threshold, initial conduction attenuation coefficient and initial superposition method of each light-emitting unit in the scene.

[0130] The central processing unit preprocesses the collected actual scene parameter information, converting it into a high-dimensional feature vector consistent with the input requirements of the correspondence model. Then, this vector is input into the correspondence model, which calculates and outputs the initial conduction threshold, initial conduction attenuation coefficient, and initial superposition method for each emitting unit.

[0131] Step S139: Based on the actual transmission and superposition effect recorded in the cross-unit transmission image of the chromaticity features, adjust the initial transmission threshold, the initial transmission attenuation coefficient and the initial superposition method. Associate the adjusted transmission threshold, transmission attenuation coefficient and cross-superposition adaptation method with the scene parameter threshold conditions that trigger their use and store them according to scene classification to generate a dynamic adaptation transmission rule set.

[0132] The initial parameters generated by the correspondence model are applied to the chroma feature transmission process. Then, chroma feature transmission images across units are acquired, and the actual transmission and superposition effect is analyzed. The actual superposition effect is compared with the standard superposition effect, and the deviation value is calculated. If the deviation between the actual superposition effect and the standard superposition effect exceeds a preset threshold, the initial transmission threshold, initial transmission attenuation coefficient, and initial superposition method are adjusted according to the deviation. For example, if the actual superposition intensity is too high, the transmission threshold is appropriately lowered or the attenuation coefficient is increased. After adjustment, the above parameters are associated with the scene parameter threshold conditions that trigger their use (such as ambient light intensity within a certain range, display workload reaching a certain value, etc.), and categorized and stored according to scene type to form a dynamically adapted transmission rule set. When the actual scene parameters meet a certain threshold condition, the system automatically calls the corresponding transmission rule to control the chroma feature transmission.

[0133] Step S140: Based on the chromaticity feature cross-unit transmission image, the chromaticity feature cross-unit transmission closed-loop link, and the dynamic adaptation transmission rule set, integrate the differentiated correction parameters of each emitting unit in different transmission directions and the cross-unit collaborative correction correlation parameters to generate a multi-dimensional coupled correction signal matrix.

[0134] First, the actual transmission feature data of each emitting unit is extracted from the cross-unit transmission image of chroma features, including initial transmission features, transmission path parameters, and superposition effects. Then, combined with the closed-loop link of cross-unit transmission of chroma features, the connection relationship and transmission efficiency of each emitting unit in different transmission directions are determined. Based on the dynamically adapted transmission rule set, rule parameters such as transmission threshold, attenuation coefficient, and superposition method in the current scene are obtained. Based on this data, differentiated correction parameters for each emitting unit in different transmission directions are calculated to correct deviations in the transmission process. Simultaneously, the cooperative relationship between emitting units is analyzed to determine cross-unit cooperative correction correlation parameters, used to coordinate the correction actions of multiple emitting units. Finally, the above differentiated correction parameters and cooperative correction correlation parameters are integrated according to a certain matrix structure to form a multi-dimensional coupled correction signal matrix. The rows and columns of this multi-dimensional coupled correction signal matrix correspond to the coordinates of the emitting units, and the matrix elements contain the correction parameters and cooperative parameters of the emitting unit in different directions.

[0135] Step S141: Extract the actual initial conduction features, actual conduction path parameters, and actual cross-over effect data of each luminescent unit from the cross-unit conduction image of the chromaticity features.

[0136] Image feature extraction algorithms are used to process the cross-unit conduction image of chromaticity features. For the body region of each emitting unit, the actual initial conduction features are extracted according to the method in step S122, including core chromaticity parameters, conduction intensity benchmark, and conduction direction preference. The extraction of actual conduction path parameters is achieved by analyzing the chromaticity data of the gap region and the indirect conduction influence region, including the path direction, length, link impedance, and conduction efficiency coefficient. The actual cross-over effect data is obtained by analyzing the chromaticity values ​​and spectral distribution after multiple conduction features are superimposed in the same region, such as the superimposed intensity, color temperature, and color coordinates. The above data are organized and stored according to the coordinates of the emitting units to form a three-dimensional data array, where each element corresponds to the actual conduction feature data of one emitting unit.

[0137] Step S142: Based on the chromaticity features, determine the forward conduction direction, reverse feedback direction, and corresponding conduction link node information of each light-emitting unit.

[0138] The chromaticity feature cross-unit conduction closed-loop link is a bidirectional network structure containing forward conduction and reverse feedback. For each emitting unit, its forward conduction direction (i.e., to which neighboring units it sends conducted features) and reverse feedback direction (i.e., from which neighboring units it receives feedback signals) are determined by querying the network's connectivity relationships. Simultaneously, the conduction link node information for each direction is recorded, including node coordinates, link weights (conduction efficiency coefficients or feedback transmission efficiency), and jump node positions. This information is used for subsequent calculations of differential correction parameters and collaborative correction correlation parameters.

[0139] Step S143: Call the conduction threshold, conduction attenuation coefficient and cross-over adaptation method corresponding to the current scene in the dynamic adaptation conduction rule set as the benchmark rule for calculating the correction parameters.

[0140] Based on the currently collected actual scene parameters, the corresponding scene category is searched in the dynamic adaptation transmission rule set, and the transmission threshold, transmission attenuation coefficient, and cross-overfitting adaptation method under that category are called. These parameters serve as the benchmark rules for calculating correction parameters, used to determine whether and how the actual transmission features need correction. For example, when the intensity of the actual initial transmission feature exceeds the transmission threshold, attenuation correction is required; when multiple transmission features are cross-overfitting, they are processed according to the cross-overfitting adaptation method.

[0141] Step S144: Compare the actual initial conduction characteristics of each light-emitting unit with the standard initial conduction characteristics in the rule set, determine the deviation of the initial conduction characteristics, and calculate the basic correction parameters of the light-emitting unit based on the deviation.

[0142] The actual initial conduction characteristics of each emitting unit are compared with the corresponding standard initial conduction characteristics in the dynamic adaptation conduction rule set. For the core chromaticity parameter, the Euclidean distance between the actual value and the standard value is calculated as the deviation; for the conduction intensity reference, the difference between the actual value and the standard value is calculated as the deviation; for the conduction direction preference, the cosine of the angle between the actual direction vector and the standard direction vector is calculated as the deviation. Then, based on the magnitude and direction of the deviation, a basic correction parameter is calculated. For example, if the actual conduction intensity reference is higher than the standard value, the basic correction parameter is a negative adjustment value used to reduce the conduction intensity; if there is a deviation between the actual direction preference and the standard direction, the basic correction parameter is a direction adjustment vector used to correct the conduction direction.

[0143] Step S145: Analyze the differences between the actual conduction path parameters of each light-emitting unit in different conduction directions and the standard conduction path parameters in the rule set, and generate differentiated path correction parameters for each conduction direction.

[0144] For each conduction direction (forward conduction and reverse feedback) of each light-emitting unit, the actual conduction path parameters (such as path length, link impedance, conduction efficiency coefficient, etc.) are compared with the standard conduction path parameters in the rule set. The differences in each parameter, such as path length difference, impedance difference, and efficiency difference, are calculated. Based on these differences and the importance weights of the conduction direction, differentiated path correction parameters are generated. For example, for the forward conduction direction, if the actual link impedance is higher than the standard value, the path correction parameter is an adjustment value to reduce the impedance; for the reverse feedback direction, if the actual conduction efficiency is lower than the standard value, the path correction parameter is an adjustment value to increase the efficiency.

[0145] Step S146: Calculate the deviation between the actual cross-over superposition effect of each light-emitting unit and the standard superposition effect in the rule set, and generate superposition correction parameters by combining the cross-over superposition adaptation method.

[0146] The actual cross-over effect data of each emitting unit's region is compared with the standard cross-over effect data in the rule set, and the deviation between the two is calculated, such as intensity deviation, color temperature deviation, and color coordinate deviation. Based on the cross-over adaptation method (e.g., linear cross-over, maximum value selection), the direction and magnitude of the deviation adjustment are determined. For example, if a linear cross-over method is used, and the actual cross-over intensity is higher than the standard value, the cross-over correction parameter is a negative coefficient used to weaken the cross-over effect; if the maximum value selection method is used, and the feature corresponding to the actual maximum value does not conform to the standard, the cross-over correction parameter is a selection weight used to adjust the weight ratio of different features.

[0147] Step S147: Identify the conduction correlation strength between each light-emitting unit and its directly adjacent and non-directly adjacent light-emitting units, and determine the weight parameters for cross-unit collaborative correction based on the correlation strength.

[0148] By analyzing the connection weights (conduction efficiency coefficient and feedback transmission efficiency) between each emitting unit in the cross-unit conduction closed-loop link of chromaticity characteristics, the conduction correlation strength between each emitting unit and its directly adjacent and non-directly adjacent emitting units is determined. The correlation strength is calculated using a weighted summation method, multiplying the connection weights of directly adjacent units by a higher coefficient (e.g., 0.7) and the connection weights of non-directly adjacent units by a lower coefficient (e.g., 0.3), and then summing the results to obtain the total correlation strength. Based on the magnitude of the correlation strength, a cooperative correction weight parameter is assigned to each adjacent unit. The greater the correlation strength, the larger the weight parameter, indicating a greater cooperative correction influence of the adjacent unit on the current emitting unit.

[0149] Step S148: Combine the basic correction parameters, the differentiated path correction parameters for different conduction directions, the superimposed correction parameters, and the cross-unit collaborative weight parameters according to the preset combination rules to form a multi-dimensional correction parameter set for each light-emitting unit.

[0150] The preset combination rules are formulated based on the importance and interrelationships of each parameter. First, the basic correction parameters are used as primary parameters, the differentiated path correction parameters and the superposition correction parameters as auxiliary parameters, and the cross-unit collaborative weighting parameters as adjustment parameters. During the combination process, the differentiated path correction parameters are first weighted and summed according to the conduction direction to obtain the comprehensive path correction parameters; then, the basic correction parameters, comprehensive path correction parameters, and superposition correction parameters are linearly combined to obtain the preliminary correction parameters; finally, the cross-unit collaborative weighting parameters are multiplied by the preliminary correction parameters to obtain the final multi-dimensional correction parameter set. This multi-dimensional correction parameter set contains correction information from multiple dimensions, reflecting the correction requirements of the luminescent unit under different conduction directions and superposition conditions.

[0151] For example, step S1481: parse the basic correction parameters, differentiated path correction parameters, superimposed correction parameters and cross-unit collaborative weight parameters, and obtain the predefined physical meaning label, effective value range definition and target object identifier for each parameter.

[0152] Each parameter is analyzed to clarify its physical meaning. For example, the "intensity correction value" in the basic correction parameters represents the adjustment amount to the conduction intensity, and the "direction correction vector" represents the adjustment direction and magnitude to the conduction direction. Simultaneously, the effective value range for each parameter is obtained; for example, the intensity correction value ranges from -1 to 1, ensuring that the parameter values ​​are within a reasonable range. The target object identifier indicates the corresponding light-emitting unit or conduction direction for that parameter; for example, a certain differentiated path correction parameter corresponds to the forward conduction direction of the light-emitting unit (x, y).

[0153] Step S1482: Use parameter normalization to convert all parameters into standardized parameters within a unified numerical range, eliminating dimensional differences and numerical range differences between different types of parameters.

[0154] The base correction parameters, differential path correction parameters, and superimposed correction parameters are all normalized using a min-max normalization method, converting their numerical range to between 0 and 1. For cross-unit collaborative weight parameters, since they already represent weights, they can be used directly. The specific formula for normalization is: Standardized parameter = (Actual parameter value - Minimum parameter value) / (Maximum parameter value - Minimum parameter value). Normalization ensures that different types of parameters have the same order of magnitude, facilitating combined calculations.

[0155] Step S1483: Based on the transmission priority in the dynamic adaptation transmission rule set, assign corresponding fusion weights to each type of correction parameter. The weight of the basic correction parameter is higher than that of the path correction parameter, the weight of the path correction parameter is higher than that of the superimposed correction parameter, and the cross-unit collaborative weight parameter is used as a global adjustment factor.

[0156] Based on the priority settings for different transmission processes in the dynamic adaptation transmission rule set, the highest fusion weight (e.g., 0.5) is assigned to the basic correction parameter, the medium weight (e.g., 0.3) to the path correction parameter, and the low weight (e.g., 0.2) to the superposition correction parameter. The cross-unit collaborative weight parameter does not participate in the fusion weight allocation, but is used as a global adjustment factor, which is multiplied on the combined parameters to adjust the intensity of collaborative correction.

[0157] Step S1484: Group the differentiated path correction parameters according to the transmission direction, and multiply the path correction parameters of each transmission direction by the corresponding fusion weight to obtain the weighted path correction parameters of that transmission direction.

[0158] The differentiated path correction parameters are grouped according to the forward conduction direction and the reverse feedback direction. For each conduction direction, all path correction parameters (such as length correction, impedance correction, efficiency correction, etc.) in that direction are multiplied by the fusion weight (0.3) of the path correction parameters to obtain the weighted value of each parameter. Then, all weighted parameters in the same conduction direction are summed to obtain the weighted path correction parameters for that conduction direction.

[0159] Step S1485: Vector superposition of the weighted path correction parameters in all directions to generate a comprehensive path correction parameter, which reflects the correction requirements in different conduction directions.

[0160] For each light-emitting unit, the weighted path correction parameters for its forward conduction direction and reverse feedback direction are superimposed as vectors. The vector superposition method involves adding the parameter components in both directions separately to obtain the individual components of the comprehensive path correction parameter. For example, if the weighted path correction parameters for the forward conduction direction are (a1, b1, c1) and those for the reverse feedback direction are (a2, b2, c2), then the comprehensive path correction parameter is (a1+a2, b1+b2, c1+c2). This comprehensive parameter takes into account the correction requirements for different conduction directions.

[0161] Step S1486: Multiply the basic correction parameters by the fusion weights, and then perform algebraic fusion with the comprehensive path correction parameters and the weighted superposition correction parameters to obtain the preliminary coupling correction parameters.

[0162] Multiplying the basic correction parameter by the fusion weight (0.5) yields the weighted basic correction parameter; multiplying the superposition correction parameter by the fusion weight (0.2) yields the weighted superposition correction parameter. Then, algebraically summing the weighted basic correction parameter, the integrated path correction parameter, and the weighted superposition correction parameter yields the preliminary coupling correction parameter. The formula for algebraic fusion is: Preliminary coupling correction parameter = Weighted basic correction parameter + Integrated path correction parameter + Weighted superposition correction parameter.

[0163] Step S1487: Apply the cross-unit collaborative weight parameter to the preliminary coupling correction parameter, adjust the contribution of the preliminary coupling correction parameter in the cross-unit collaborative correction according to the cross-unit collaborative weight parameter, and generate the intermediate coupling parameter.

[0164] The cross-cell collaborative weighting parameter is a coefficient related to the correlation strength of adjacent emitting cells, with a value between 0 and 1. The intermediate coupling parameter is obtained by multiplying the initial coupling correction parameter by the cross-cell collaborative weighting parameter. That is, intermediate coupling parameter = initial coupling correction parameter × cross-cell collaborative weighting parameter. In this way, the contribution of the current emitting cell's correction parameter to the cross-cell collaborative correction is adjusted; the greater the correlation strength, the higher the contribution.

[0165] Step S1488: Analyze the cooperative compatibility between the intermediate coupling parameter and the correction parameters of adjacent light-emitting units, and fine-tune the value of the intermediate coupling parameter based on the cross-superposition adaptation method in the dynamic adaptation conduction rule.

[0166] Compare the intermediate coupling parameters with the correction parameters of adjacent light-emitting units to analyze whether there are any conflicts or incompatibilities. For example, if the intermediate coupling parameters of the current light-emitting unit require enhancing the conduction intensity in a certain direction, while the correction parameters of adjacent units require weakening the conduction intensity in that direction, coordination is required. Based on the cross-superposition adaptation method in the dynamic adaptation conduction rules, such as conflict resolution, the value of the intermediate coupling parameters is fine-tuned. For example, by averaging, the average value of the current parameter and the adjacent parameters can be used as the fine-tuned intermediate coupling parameters.

[0167] Step S1489: Apply a moving average filter to smooth the fine-tuned intermediate coupling parameters, perform a correction fluctuation suppression operation caused by parameter mutation, classify the smoothed intermediate coupling parameters according to the logic of basic correction, path correction, superposition correction and collaborative correction to form a structured parameter combination, remove parameters with duplicate values ​​and invalid values ​​from the structured parameter combination, and finally form a multi-dimensional correction parameter set for each light-emitting unit.

[0168] The window size of the moving average filter is set to 3, meaning that the current intermediate coupling parameter is averaged with the parameters from the previous two time steps to obtain the smoothed parameter value. Smoothing suppresses correction fluctuations caused by parameter abrupt changes, making the correction process more stable. Then, the smoothed intermediate coupling parameters are categorized according to the logic of basic correction, path correction, superposition correction, and collaborative correction, forming a structured parameter combination. This combination is checked for parameters with duplicate values ​​(e.g., parameters of different correction types but the same value) and invalid values ​​(e.g., parameters whose values ​​exceed the valid range), and these parameters are removed. The final structured parameter combination is the multi-dimensional correction parameter set for each luminescent unit.

[0169] Step S149: Check whether the differences between path correction parameters in different transmission directions in the multi-dimensional correction parameter set exceed the preset threshold. If they exceed the preset threshold, recalculate them according to the cross-overfitting adaptation method and perform parameter co-fitting operation.

[0170] The preset threshold is set based on the color uniformity requirements of the display screen. For example, a difference exceeding 0.2 between path correction parameters in different transmission directions is considered excessive. The multi-dimensional correction parameter set for each light-emitting unit is checked, comparing the path correction parameters in the forward transmission direction and the reverse feedback direction. If the difference exceeds the preset threshold, it indicates an inconsistency in the correction parameters across different directions, requiring recalculation based on the cross-over adaptation method in the dynamic adaptation transmission rules. For example, if the cross-over adaptation method involves averaging, the average of the path correction parameters in the two directions is used as the new parameter value, and parameter co-adaptation is performed to ensure that the correction parameters in different directions are coordinated.

[0171] Step S1410: Arrange the multi-dimensional correction parameter sets of all light-emitting units in a matrix according to the physical arrangement order of the light-emitting units and the node order of the closed-loop transmission link to construct the initial correction signal matrix.

[0172] The physical arrangement of the light-emitting units follows the rows and columns of the display screen, arranged sequentially from the top left to the bottom right. The node order of the closed-loop transmission link is determined according to the connection relationship of the link, ensuring that the rows and columns of the matrix can reflect the transmission relationship between the nodes. The multi-dimensional correction parameter set of each light-emitting unit is used as an element of the matrix, arranged in the above order to construct the initial correction signal matrix. The number of rows and columns of the matrix is ​​equal to the total number of light-emitting units, and each element is a vector containing multiple correction parameters.

[0173] Step S1411: Based on the dynamic adjustment mechanism in the dynamic adaptation transmission rule set, optimize the parameters in the initial correction signal matrix for scene adaptation, and perform coupling coordination adaptation operation of the parameters between different transmission link nodes.

[0174] The dynamic adjustment mechanism adaptively adjusts the correction parameters based on changes in current scene parameters. For example, in scenarios with increased ambient light intensity, the conduction threshold needs to be appropriately increased to avoid excessive conduction interference. Based on the correspondence in the dynamic adaptation conduction rule set, each parameter in the initial correction signal matrix is ​​optimized for scene adaptation. Simultaneously, the parameter coupling relationships between different conduction link nodes are analyzed to ensure that the correction parameters of adjacent nodes are coordinated, avoiding conflicting correction commands. For instance, if a node's correction parameters require enhanced conduction, while its adjacent nodes require weakened conduction, the parameters of both need to be adjusted to achieve a balance in the overall conduction effect, performing a parameter coupling coordination adaptation operation.

[0175] Step S1412: Transmit the optimized parameter matrix between the transmission link nodes, and correct the parameter values ​​according to the feedback between the nodes to form the final multi-dimensional coupling correction signal matrix.

[0176] The scene-adapted and optimized parameter matrix is ​​transmitted between nodes via a closed-loop link for cross-unit transmission of chromaticity features. Each node receives parameter information from its neighboring nodes and compares and merges it with its own parameters. Based on feedback information between nodes, such as parameter differences and transmission effects, the node corrects its own parameter values. For example, if the parameters of node A differ significantly from those of node B, and feedback indicates poor transmission effects, the corrected parameters are derived from the average of the two. After multiple rounds of transmission and correction, the parameter matrix gradually converges, ultimately forming a stable multi-dimensional coupled correction signal matrix.

[0177] Step S150: In the chromaticity feature cross-unit transmission closed-loop link, the transmission of the multi-dimensional coupling correction signal matrix and the process of driving all light-emitting units to synchronously adjust the chromaticity output according to the multi-dimensional coupling correction signal matrix are executed cyclically. After each adjustment, the chromaticity feature cross-unit transmission image is re-acquired to update the multi-dimensional coupling correction signal matrix, and iterative correction is performed until the preset chromaticity uniformity standard is met, thereby realizing the chromaticity uniformity correction of the Miniled display screen.

[0178] First, the multi-dimensional coupling correction signal matrix is ​​transmitted to the drive control unit of each light-emitting unit through a closed-loop link for chromaticity feature transmission across units. The drive control unit adjusts the chromaticity output of the light-emitting unit according to the received correction parameters, including intensity, spectrum, and stability. After adjustment, a spectral imaging device is used to re-acquire the chromaticity feature transmission image across units, and new transmission feature data is extracted. Based on the new transmission feature data, the multi-dimensional coupling correction signal matrix is ​​updated according to the method in step S140. Then, the matrix transmission and chromaticity output adjustment are performed again, and this process is repeated iteratively. After each iteration, the chromaticity uniformity of the display screen is checked to see if it meets the preset standard, such as the color coordinate deviation of all light-emitting units being within ±0.002 and the brightness deviation being within ±5%. If the standard is met, the correction stops; otherwise, the next iteration continues until the preset standard is reached.

[0179] Step S151: Analyze the multi-dimensional coupling correction signal matrix and extract the correction parameter subsets of each light-emitting unit in the forward conduction direction and the reverse feedback direction.

[0180] The multi-dimensional coupling correction signal matrix is ​​a large matrix containing correction parameters for all light-emitting units. By indexing the matrix, the row or column corresponding to each light-emitting unit is located, and the multi-dimensional correction parameter set for that unit is extracted. Then, based on the physical meaning labels of the correction parameters, the parameter set is divided into subsets for the forward propagation direction and the reverse feedback direction. The subset for the forward propagation direction includes path correction parameters, superimposed correction parameters, etc., in that direction; the subset for the reverse feedback direction includes the corresponding path correction parameters, feedback response parameters, etc. The extracted correction parameter subsets are then sent to the corresponding drive control unit.

[0181] Step S152: Based on the node distribution of the cross-unit transmission closed-loop link of the chromaticity features, determine the transmission path and transmission priority of each subset of correction parameters.

[0182] The node distribution of the chromaticity feature cross-unit conduction closed-loop link reflects the connection relationship between the emitting units. For each subset of correction parameters, a transmission path from the current node to the target node is found in the link according to its corresponding conduction direction (forward or reverse). The transmission priority is determined based on the link's conduction efficiency coefficient and feedback transmission efficiency; paths with higher efficiency have higher priority. For example, in the forward conduction direction, paths with higher conduction efficiency coefficients have higher priority; in the reverse feedback direction, paths with higher feedback transmission efficiency have higher priority. After determining the transmission paths and priorities, a parameter transmission schedule table is generated to guide the parameter transmission process.

[0183] Step S153: Monitor the transmission delay of the closed-loop link in real time, calculate and attach a transmission delay compensation timestamp for each subset of correction parameters based on the transmission delay data, and perform parameter synchronization transmission scheduling operation.

[0184] The transmission time of the correction parameter subset from the sending node to the receiving node is recorded using timestamps, and the transmission delay is calculated. Transmission delay data is collected in real time by the detection nodes in the closed-loop link. Based on the magnitude of the transmission delay, a compensation timestamp is calculated for each correction parameter subset, which is the transmission time plus the transmission delay, ensuring that the receiving node receives the parameters at the expected time. Based on the compensation timestamps and transmission priorities, a parameter synchronization transmission scheduling operation is performed to coordinate the transmission order and time of multiple parameter subsets, avoiding transmission conflicts and data loss, and ensuring that all emitting units can receive the correction parameters simultaneously.

[0185] Step S154: According to the transmission priority and the compensated timing, the forward conduction correction parameters of each light-emitting unit are transmitted to the corresponding conduction object along the forward conduction path of the closed-loop link, and the reverse feedback correction parameters are transmitted to the feedback starting unit along the reverse feedback path.

[0186] According to the parameter transmission schedule, each subset of correction parameters is sent sequentially in descending order of transmission priority. During transmission, the compensated timing is strictly followed to ensure that each parameter subset arrives at the target node at the correct time. Forward-propagating correction parameters are transmitted along the forward propagation path of the closed-loop link to directly adjacent and non-directly adjacent propagation objects; backward-feedback correction parameters are transmitted back along the backward-feedback path to the starting unit that sent the original propagation characteristics. During transmission, a verification mechanism is employed to ensure data integrity and accuracy, such as using a CRC checksum to verify the parameter data.

[0187] Step S155: The drive control unit of each light-emitting unit receives its own corresponding subset of correction parameters and the collaborative correction parameters from other units, and performs weighted calculation according to the algorithm corresponding to the cross-superposition adaptation method in the dynamic adaptation transmission rule. Based on the weighted correction parameters, the chromaticity output control circuit inside the drive control unit is adjusted to change the chromaticity output intensity, output spectrum and output stability parameters of the light-emitting unit.

[0188] After receiving its own subset of correction parameters and the cooperative correction parameters from neighboring units, the drive control unit first verifies these parameters to ensure data validity. Then, according to the cross-superposition adaptation method in the dynamic adaptation transmission rules, such as a linear superposition algorithm, it performs a weighted calculation on all parameters. The weights in the weighted calculation are determined based on the cooperative correction associated parameters, with the weight of the user's own parameters being higher than the weight of the cooperative parameters. The calculated comprehensive correction parameters are used to adjust the chromaticity output control circuit within the drive control unit.

[0189] For example, in step S1551: the weighted calculation of the correction parameters is transmitted to the parameter parsing module of the drive control unit, and the independent control parameters for controlling the chromaticity output intensity, output spectrum and output stability are separated according to the preset parameter encoding rules.

[0190] The parameter parsing module decodes the received integrated correction parameters and, according to a preset parameter encoding rule, breaks them down into independent control parameters that control the chroma output intensity, output spectrum, and output stability. The parameter encoding rule specifies the start bit, length, and data format of each parameter; for example, the first 16 bits represent the intensity control parameter, the middle 24 bits represent the spectrum control parameter, and the last 8 bits represent the stability control parameter. The parsed independent control parameters are then sent to their respective control circuits.

[0191] Step S1552: Verify whether the data format and value range of the split independent control parameters conform to the interface protocol of the drive control unit. Transmit the chromaticity output intensity control parameters to the intensity control circuit of the drive control unit. The intensity control circuit adjusts the internal power amplification module to change the amplitude of the drive current output to the light-emitting component, thereby adjusting the chromaticity output intensity of the light-emitting unit.

[0192] The parameter parsing module verifies the format and range of the split, independent control parameters to ensure that the data format is correct (e.g., binary, decimal) and the values ​​are within the preset valid range (e.g., the range of intensity control parameters is 0-1000mA). After verification, the chromaticity output intensity control parameters are transmitted to the intensity control circuit. The power amplifier module inside the intensity control circuit adjusts the amplitude of the output current according to this parameter. The larger the current amplitude, the higher the luminous intensity of the light-emitting component, thereby adjusting the chromaticity output intensity.

[0193] Step S1553: The output spectrum control parameters are sent to the spectrum adjustment circuit. The spectrum adjustment circuit changes the frequency distribution of the driving signal by switching the internal filter module channel, so as to achieve precise adjustment of the output spectrum of the light-emitting unit and make the spectrum meet the calibration requirements.

[0194] The output spectrum control parameters are sent to the spectrum adjustment circuit. These parameters contain characteristic information of the target spectrum, such as the intensity ratio of each frequency component. The spectrum adjustment circuit internally contains multiple filter module channels with different frequency characteristics, switching to the appropriate channel according to the control parameters. Different filter channels filter and adjust the frequency components of the drive signal, ensuring that the frequency distribution of the drive signal output to the light-emitting component meets the correction requirements, thereby changing the output spectrum of the light-emitting unit.

[0195] Step S1554: Input the output stability control parameters to the stability control circuit. The stability control circuit starts and monitors the chromaticity output signal of the light-emitting unit in real time. When the fluctuation amplitude exceeds the preset threshold, adjust the pulse width and frequency of the drive signal to perform output fluctuation suppression operation.

[0196] Output stability control parameters are input to the stability control circuit, which sets the allowable threshold for output signal fluctuations. The stability control circuit activates its internal monitoring module to acquire the chromaticity output signal of the light-emitting unit in real time and calculate its fluctuation amplitude. When the fluctuation amplitude exceeds the preset threshold, the stability control circuit adjusts the pulse width and frequency of the drive signal; for example, by increasing the pulse width to improve stability or adjusting the frequency to avoid interference frequencies, it performs output fluctuation suppression operations to ensure the stability of the chromaticity output.

[0197] Step S1555: The central timing controller sends a synchronous trigger signal to the intensity control circuit, the spectrum adjustment circuit, and the stability control circuit, and monitors the response of each circuit. If the response times out, the triggering of the next cycle is paused, and the synchronous scheduling operation of the adjustment actions of each circuit is executed, and the adjustment timing difference adaptation operation is performed.

[0198] The central timing controller sends synchronous trigger signals to the three control circuits at preset time intervals to ensure they begin adjustment actions simultaneously. Upon receiving the trigger signal, each circuit executes its corresponding adjustment operation and returns a response signal to the central timing controller. The central timing controller monitors the response time of each circuit. If a circuit times out, it indicates a potential fault or delay. In this case, the transmission of trigger signals for the next cycle is paused until the fault or delay is resolved. Simultaneously, based on the differences in response times among the circuits, the controller performs timing adjustment adjustments, such as adjusting the trigger signal transmission time, to ensure that the adjustment actions of each circuit are completed synchronously.

[0199] Step S1556: The monitoring module inside the drive control unit collects the output signal parameters of each circuit in real time, including the drive current amplitude, signal frequency distribution and pulse stability data, to form real-time control feedback data.

[0200] The monitoring module has a built-in current sensor, frequency analyzer, and pulse monitoring circuit. It collects the driving current amplitude output by the intensity control circuit, the signal frequency distribution output by the spectrum adjustment circuit, and the pulse stability data (such as pulse width, frequency, duty cycle, etc.) output by the stability control circuit in real time. The above data is integrated into real-time control feedback data and sent to the central processing unit for evaluation of the correction effect.

[0201] Step S1557: Compare the real-time control feedback data with the expected control target corresponding to the weighted correction parameters to generate a control deviation signal, which reflects the gap between the actual control effect and the expected target.

[0202] The central processing unit compares the real-time control feedback data with the expected control target corresponding to the weighted correction parameters. For example, if the expected drive current amplitude is 500mA and the actual current amplitude is 480mA, the control deviation signal is -20mA. By calculating the deviation values ​​of various parameters, a comprehensive control deviation signal is generated, which quantifies the gap between the actual control effect and the expected target.

[0203] Step S1558: Based on the control deviation signal, adjust the control parameters of each control circuit, correct the output characteristics of the drive signal, and narrow the gap between the actual control effect and the expected target.

[0204] Based on the magnitude and direction of the control deviation signal, the central processing unit sends parameter adjustment commands to each control circuit. For example, if the current amplitude deviation is -20mA, a command to increase the current by 20mA is sent to the intensity control circuit; if the frequency distribution deviation is large, a command to adjust the filter channel is sent to the spectrum adjustment circuit. Each control circuit corrects its own control parameters according to the adjustment commands, thereby correcting the output characteristics of the drive signal, so that the actual control effect gradually approaches the expected target.

[0205] Step S1559: Continuously repeat the process of parameter analysis, parameter verification, circuit adjustment, real-time monitoring, deviation comparison and parameter correction until the driving current amplitude, signal frequency distribution and pulse stability data enter the target range specified by the correction parameters, and perform the adaptation operation of the chromaticity output intensity, output spectrum and output stability parameters of the light-emitting unit with the target range corresponding to the correction parameters.

[0206] The drive control unit continuously repeats the above process of parameter analysis, verification, adjustment, monitoring, comparison, and correction, forming a closed-loop control circuit. In each cycle, the control parameters are continuously adjusted based on feedback data until the drive current amplitude, signal frequency distribution, and pulse stability data all fall within the target range specified by the calibration parameters. At this point, an adaptation operation is performed to fix each control parameter, stabilizing the chromaticity output of the light-emitting unit within the target range.

[0207] Step S1560: Pass the stabilized output signal through a low-pass filter and continuously output the filtered and stabilized drive signal to the light-emitting component of the light-emitting unit, driving the light-emitting component to output stably according to the corrected chromaticity parameters.

[0208] The stabilized drive signal may contain high-frequency noise. This noise is filtered by a low-pass filter with a cutoff frequency of 1MHz to remove the high-frequency noise, resulting in a smoother and more stable drive signal. The filtered drive signal is continuously output to the light-emitting component of the light-emitting unit, driving the light-emitting component to emit light stably according to the corrected chromaticity parameters (intensity, spectrum, stability), thus achieving chromaticity uniformity correction.

[0209] Step S156: Acquire the real-time chromaticity feature cross-unit transmission image of all light-emitting units after adjustment, and extract the adjusted real-time initial transmission features, real-time transmission path parameters and real-time cross-over effect.

[0210] After the chromaticity output of the light-emitting unit stabilizes, a spectral imaging device is used to re-acquire the chromaticity feature cross-unit transmission image of the entire display screen, i.e., the real-time chromaticity feature cross-unit transmission image. Using the same method as in step S141, the real-time initial transmission characteristics, real-time transmission path parameters, and real-time cross-over effect data of each light-emitting unit are extracted from this image as the basis for evaluating the correction effect and updating the correction parameters.

[0211] Step S157: Compare the adjusted real-time data with the standard data in the dynamic adaptation conduction rule set to generate correction effect deviation information for each light-emitting unit.

[0212] The extracted real-time initial conduction features, real-time conduction path parameters, and real-time cross-over effect data are compared item by item with the standard data in the corresponding scenario of the dynamic adaptation conduction rule set. Deviation values ​​for each data point are calculated, such as the Euclidean distance between the real-time initial conduction features and the standard initial conduction features, the difference between the real-time conduction path parameters and the standard parameters, and the percentage deviation between the real-time overlay effect and the standard overlay effect. These deviation values ​​are then integrated to generate correction effect deviation information for each luminescent unit, reflecting the gap between the current correction effect and the ideal state.

[0213] Step S158: Based on the correction effect deviation information, correct the corresponding parameters in the multi-dimensional coupled correction signal matrix to generate an optimized correction signal matrix.

[0214] Based on the magnitude and direction of the deviation information in the correction effect, the correction parameters of the corresponding light-emitting units in the multi-dimensional coupled correction signal matrix are adjusted. For example, if the real-time initial conduction intensity of a certain light-emitting unit is lower than the standard value and the deviation information is negative, the intensity correction value in the basic correction parameters of that unit is increased; if the real-time superposition effect deviation is large, the superposition correction parameters are adjusted. The corrected parameter matrix is ​​the optimized correction signal matrix, which is used for the next round of correction iteration.

[0215] Step S159: The optimized correction signal matrix is ​​cyclically transmitted again through the chromaticity feature cross-unit transmission closed loop link, repeating the above steps of parameter reception, fusion, drive adjustment, real-time acquisition, deviation comparison and parameter correction.

[0216] The optimized correction signal matrix is ​​then transmitted, received, fused, driven, adjusted, image acquired, compared, and corrected again following steps S151 to S158. Each cycle further optimizes the correction parameters, continuously improving the color uniformity of the display screen.

[0217] Step S1510: Continuously perform cyclic transmission of the correction signal matrix, parameter optimization and chromaticity output adjustment until the deviation between the real-time initial conduction characteristics, real-time conduction path parameters and real-time cross-over effect of all light-emitting units and the standard data in the dynamic adaptation conduction rules is less than the preset threshold, thereby achieving chromaticity uniformity correction of the Miniled display screen.

[0218] The process of iteratively transferring and optimizing parameters is repeated, with each iteration checking whether the deviation is less than a preset threshold. The preset threshold is set according to the application requirements of the display screen, such as a color coordinate deviation threshold of ±0.002, a brightness deviation threshold of ±5%, and a transmission path parameter deviation threshold of ±10%. When the deviations between the real-time data of all light-emitting units and the standard data are all less than the preset thresholds, it indicates that the color uniformity of the display screen has met the requirements, and the calibration process stops, completing the color uniformity calibration of the Miniled display screen.

[0219] Based on the same inventive concept, please refer to Figure 2 The diagram shows a schematic block diagram of the structure of a Miniled display screen chromaticity uniformity correction device 100 based on image feature matching provided in an embodiment of this application. The Miniled display screen chromaticity uniformity correction device 100 based on image feature matching may include a communication unit 110, a machine-readable storage medium 120, and a processor 130.

[0220] In this embodiment, the machine-readable storage medium 120 can also be integrated into the processor 130 and can communicate and interact with external systems through the communication unit 110. The machine-readable storage medium 120 stores machine-executable instructions for executing the scheme of this application, and the processor 130 executes the machine-executable instructions stored in the machine-readable storage medium 120 to implement the image feature matching-based color uniformity correction method for minimized displays provided in the aforementioned method embodiments.

[0221] Furthermore, embodiments of the present invention also provide a display screen, including a display panel on which a plurality of minimized light-emitting units are arranged; and a minimized display screen color uniformity correction device 100 based on image feature matching, wherein the minimized display screen color uniformity correction device 100 based on image feature matching is electrically connected to the display panel and is configured to perform the minimized display screen color uniformity correction method based on image feature matching of the above embodiments.

[0222] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for correcting the chromaticity uniformity of a minimized display screen based on image feature matching, characterized in that, The method includes: Acquire cross-unit transmission images of chromaticity features of a Miniled display screen. These cross-unit transmission images record the transmission and diffusion state of chromaticity features of all light-emitting units to non-directly adjacent light-emitting units through the gaps between adjacent units, as well as the superposition effect of cross-transmission of multiple units. Based on the chromaticity feature transmission paths of directly adjacent and non-directly adjacent light-emitting units, a closed-loop structure containing forward transmission and reverse feedback is formed by connecting them in series, thus constructing a closed-loop link for chromaticity feature transmission across units. Based on the chromaticity feature transmission characteristics under different working scenarios, the chromaticity feature transmission threshold, transmission attenuation coefficient and cross-superposition adaptation method of each light-emitting unit are defined to generate a dynamic adaptation transmission rule set. Based on the chromaticity feature cross-unit transmission image, the chromaticity feature cross-unit transmission closed-loop link, and the dynamic adaptation transmission rule set, the differentiated correction parameters of each luminescent unit in different transmission directions and the cross-unit collaborative correction correlation parameters are integrated to generate a multi-dimensional coupled correction signal matrix. In the chromaticity feature cross-unit transmission closed-loop link, the transmission of the multi-dimensional coupling correction signal matrix and the process of driving all light-emitting units to synchronously adjust the chromaticity output according to the multi-dimensional coupling correction signal matrix are executed cyclically. After each adjustment, the chromaticity feature cross-unit transmission image is re-acquired to update the multi-dimensional coupling correction signal matrix, and iterative correction is performed until the preset chromaticity uniformity standard is met, thereby realizing the chromaticity uniformity correction of the Miniled display screen.

2. The method for correcting the color uniformity of a minimized display screen based on image feature matching according to claim 1, characterized in that, The chromaticity feature transmission paths based on directly adjacent and non-directly adjacent light-emitting units are connected in series to form a closed-loop structure containing forward transmission and reverse feedback, constructing a cross-unit transmission closed-loop link for chromaticity features, including: The chromaticity feature cross-unit transmission image is segmented into light-emitting units and gap regions. The body region of each light-emitting unit, the gap region between adjacent units, and the indirect transmission influence region of non-directly adjacent units are separated. In the metadata of each region generated after the segmentation process, the chromaticity feature transmission information associated with that region is recorded. The initial conduction characteristics of the body region of each light-emitting unit are extracted. The initial conduction characteristics reflect the conduction initiation attribute, conduction intensity reference, and initial conduction direction preference of the chromaticity characteristics of the light-emitting unit. Identify the directly adjacent and non-directly adjacent light-emitting units of each light-emitting unit, and record the hierarchical relationship and physical location distribution of the directly and indirectly transmitted objects; The forward conduction trajectory from the light-emitting unit body region through the gap region to the directly adjacent light-emitting unit is traced based on the initial conduction characteristics, and the direction, length and conduction resistance distribution of the direct conduction path are determined. The indirect conduction trajectory of the initial conduction characteristics is analyzed from the direct adjacent light-emitting unit to the non-directly adjacent light-emitting unit, and the jump nodes, conduction delay and characteristic attenuation characteristics of the indirect conduction path are determined. Calculate the link impedance of each direct and indirect conduction path, and assign a conduction efficiency coefficient to each path based on the link impedance to form a set of conduction paths after impedance matching. Based on the set of conduction paths and jump nodes after impedance matching, a forward conduction link network covering all cross-unit conduction paths of light-emitting units is constructed. Extract the conduction characteristic feedback information received by each light-emitting unit from other units, and determine the start point, end point and feedback transmission efficiency of the reverse feedback path; The starting and ending points of the reverse feedback path are mapped to the corresponding nodes in the forward transmission link network to construct a bidirectional link framework that includes forward transmission and reverse feedback. In a bidirectional link framework, a transmission feature detection node is deployed, which captures in real time the intersection and superposition effect of positive transmission features and negative feedback features. Based on the capture results of the detection nodes, the transmission priority parameters and feedback response rate parameters of the link are updated to form a closed-loop link for cross-unit transmission of chromaticity features.

3. The method for correcting the color uniformity of a minimized display screen based on image feature matching according to claim 1, characterized in that, The chromaticity feature transmission characteristics based on different working scenarios are defined as follows: The chromaticity characteristic transmission threshold, transmission attenuation coefficient, and cross-superposition adaptation method of each emitting unit are used to generate a dynamic adaptation transmission rule set, including: Standard chromaticity feature transmission images of a Miniled display screen are acquired under various preset working scenarios. The standard chromaticity feature transmission images record the cross-unit transmission characteristics of the ideal chromaticity uniformity state under different scenarios. For the standard chromaticity feature transmission image in each scenario, image segmentation and feature recognition algorithms are applied to obtain the standard initial transmission features, standard transmission paths and standard superposition effects of all luminous units in that scenario. Analyze the variation characteristics of the standard initial conduction characteristics on different conduction paths under each scenario, and define the standard conduction threshold of each light-emitting unit under the scenario. The standard conduction threshold is the critical value of conduction intensity corresponding to chromaticity uniformity. Calculate the standard attenuation magnitude of the standard conduction feature in each scenario on the direct and indirect conduction paths, and define the standard conduction attenuation coefficient for different paths based on the attenuation magnitude. We analyze the standard patterns of cross-superposition of the conduction features of multiple light-emitting units in each scenario, and define the feature complementarity rules and conflict resolution methods in the superposition process. A scene parameter is assigned a higher weight coefficient than other scene parameters for ambient lighting conditions, display workload, and target display content type. Based on the scene parameters after weight allocation, a correspondence model is established between scene parameters and standard conduction threshold, standard conduction attenuation coefficient, and standard superposition method. The correspondence model reflects the influence of different scene parameters on the conduction rule. Collect scenario parameter information in actual applications, including ambient lighting conditions, display workload, and key factors affecting color transmission, such as the type of target display content. Input the actual scene parameter information into the corresponding relationship model to generate the initial conduction threshold, initial conduction attenuation coefficient and initial superposition method of each light-emitting unit in the scene; Based on the actual transmission and superposition effect recorded in the cross-unit transmission image of the chromaticity features, the initial transmission threshold, initial transmission attenuation coefficient and initial superposition method are adjusted. The adjusted transmission threshold, transmission attenuation coefficient and cross superposition adaptation method are associated with the scene parameter threshold conditions that trigger their use and stored according to scene classification to generate a dynamic adaptation transmission rule set.

4. The method for correcting the color uniformity of a minimized display screen based on image feature matching according to claim 1, characterized in that, The method, based on the chromaticity feature cross-unit transmission image, the chromaticity feature cross-unit transmission closed-loop link, and the dynamic adaptation transmission rule set, integrates the differentiated correction parameters of each emitting unit in different transmission directions and the cross-unit collaborative correction correlation parameters to generate a multi-dimensional coupled correction signal matrix, including: Extract the actual initial conduction features, actual conduction path parameters, and actual cross-over effect data of each luminescent unit from the cross-unit conduction image of the chromaticity features; Based on the chromaticity characteristics of the cross-unit conduction closed-loop link, the forward conduction direction, reverse feedback direction and corresponding conduction link node information of each light-emitting unit are determined; The current scene's conduction threshold, conduction attenuation coefficient, and cross-over adaptation method are called from the dynamic adaptation conduction rule set as the benchmark rules for calculating the correction parameters. By comparing the actual initial conduction characteristics of each light-emitting unit with the standard initial conduction characteristics in the rule set, the deviation of the initial conduction characteristics is determined, and the basic correction parameters of the light-emitting unit are calculated based on the deviation. The differences between the actual conduction path parameters of each light-emitting unit in different conduction directions and the standard conduction path parameters in the rule set are analyzed, and differentiated path correction parameters are generated for each conduction direction. Calculate the deviation between the actual cross-over superposition effect of each light-emitting unit and the standard superposition effect in the rule set, and generate superposition correction parameters by combining the cross-over superposition adaptation method; Identify the conduction correlation strength between each light-emitting unit and its directly adjacent and non-directly adjacent light-emitting units, and determine the weighting parameters for cross-unit collaborative correction based on the correlation strength; The basic correction parameters, the differentiated path correction parameters for different conduction directions, the superimposed correction parameters, and the cross-unit collaborative weight parameters are combined according to the preset combination rules to form a multi-dimensional correction parameter set for each light-emitting unit. Check whether the differences between path correction parameters in different transmission directions in the multi-dimensional correction parameter set exceed the preset threshold. If they exceed the preset threshold, recalculate them according to the cross-over superposition adaptation method and perform parameter co-adaptation operation. According to the physical arrangement order of the light-emitting units and the node order of the closed-loop transmission link, the multi-dimensional correction parameter sets of all light-emitting units are arranged in a matrix to construct the initial correction signal matrix; Based on the dynamic adjustment mechanism in the dynamic adaptation transmission rule set, the parameters in the initial correction signal matrix are optimized for scene adaptation, and the coupling coordination adaptation operation of the parameters is performed between different transmission link nodes. The optimized parameter matrix is ​​transmitted between the nodes of the transmission link, and the parameter values ​​are corrected based on the feedback between the nodes to form the final multi-dimensional coupling correction signal matrix.

5. The method for correcting the chromaticity uniformity of a minimized display screen based on image feature matching according to claim 1, characterized in that, In the chromaticity feature cross-unit transmission closed-loop link, the process of cyclically transmitting the multi-dimensional coupling correction signal matrix and driving all light-emitting units to synchronously adjust the chromaticity output according to the multi-dimensional coupling correction signal matrix is ​​executed. After each adjustment, the chromaticity feature cross-unit transmission image is re-acquired to update the multi-dimensional coupling correction signal matrix, and iterative correction is performed until a preset chromaticity uniformity standard is met, thereby realizing the chromaticity uniformity correction of the Miniled display screen, including: The multi-dimensional coupling correction signal matrix is ​​analyzed, and a subset of correction parameters for each light-emitting unit in the forward conduction direction and the reverse feedback direction is extracted. Based on the node distribution of the cross-unit transmission closed-loop link of the chromaticity features, the transmission path and transmission priority of each subset of correction parameters are determined. Real-time monitoring of the transmission delay of the closed-loop link; calculation of transmission delay compensation timestamps for each subset of correction parameters based on the transmission delay data; and execution of parameter synchronization transmission scheduling operations. According to the transmission priority and the compensated timing, the forward conduction correction parameters of each light-emitting unit are transmitted to the corresponding conduction object along the forward conduction path of the closed-loop link, while the reverse feedback correction parameters are transmitted to the feedback starting unit along the reverse feedback path. Each light-emitting unit's drive control unit receives its own corresponding subset of correction parameters and collaborative correction parameters from other units, and performs weighted calculations according to the algorithm corresponding to the cross-superposition adaptation method in the dynamic adaptation transmission rule. Based on the weighted correction parameters, the chromaticity output control circuit inside the drive control unit is adjusted to change the chromaticity output intensity, output spectrum, and output stability parameters of the light-emitting unit. Collect real-time chromaticity feature cross-unit transmission images of all light-emitting units after adjustment, and extract the adjusted real-time initial transmission features, real-time transmission path parameters, and real-time cross-over effect. The adjusted real-time data is compared with the standard data in the dynamic adaptation conduction rule set to generate correction effect deviation information for each light-emitting unit; Based on the correction effect deviation information, the corresponding parameters in the multi-dimensional coupled correction signal matrix are corrected to generate an optimized correction signal matrix; The optimized correction signal matrix is ​​then cyclically transmitted through the chromaticity feature cross-unit transmission closed-loop link, repeating the above steps of parameter reception, fusion, drive adjustment, real-time acquisition, deviation comparison and parameter correction. The correction signal matrix is ​​continuously cyclically transmitted, parameters are optimized, and chromaticity output is adjusted until the deviation between the real-time initial conduction characteristics, real-time conduction path parameters, and real-time cross-over effect of all light-emitting units and the standard data in the dynamic adaptation conduction rules is less than the preset threshold, thereby achieving chromaticity uniformity correction of the Miniled display screen.

6. The method for correcting the chromaticity uniformity of a minimized display screen based on image feature matching according to claim 2, characterized in that, The initial conduction characteristics of each luminescent unit's body region are extracted. These initial conduction characteristics reflect the conduction initiation attributes, conduction intensity benchmark, and initial conduction direction preference of the luminescent unit's chromaticity characteristics, including: A pixel-level fine scan is performed on the body area of ​​each light-emitting unit to collect the chromaticity values, brightness distribution information and chromaticity correlation data of all pixels in the body area; Using the geometric center of the light-emitting unit body area as a reference point, multiple concentric annular regions are divided, and each annular region contains pixels at the same radial distance from the reference point; Analyze the concentration of chromaticity values ​​of pixels within each annular region, and extract the dominant chromaticity attribute of each annular region. The dominant chromaticity attribute is the most representative chromaticity feature within that annular region. Calculate the gradient of dominant chromaticity attribute changes from the central annular region to the edge annular region to determine the distribution characteristics and diffusion initiation characteristics of chromaticity features within the body region; The core chromaticity parameters of the central annular region are extracted. These core chromaticity parameters are the essential manifestation of the chromaticity characteristics of the luminescent unit and serve as the basis for calculating the conduction intensity benchmark. Based on the core chromaticity parameters and the dominant chromaticity attributes of the edge annular region, an initial transmission intensity benchmark value is calculated, which reflects the initial intensity level of chromaticity features transmitted outward. Analyze the spatial gradient distribution of pixel chromaticity values ​​within the body region to determine the difference in diffusion potential energy of chromaticity features in different directions, where the difference in diffusion potential energy reflects the initial conduction direction preference. An integration operation is performed on the core chromaticity parameters, the initial transmission intensity benchmark value, the initial transmission direction preference, and the gradient of the dominant chromaticity attribute change in the annular region to form the basic components of the initial transmission characteristics; Remove redundant information from the basic components that is irrelevant to cross-unit transmission, and retain the key elements that can accurately reflect the transmission initiation attributes; Following the logical order of core chromaticity parameters, conduction intensity benchmark, conduction direction preference, and regional variation gradient, key elements are structured and organized to form initial conduction features. The initial conduction feature data after structured organization are then normalized and principal component analysis is performed sequentially. Principal components with variance contribution rates exceeding a preset threshold are retained and standardized and scaled. Feature recognition enhancement and stability adaptation operations are then performed.

7. The method for correcting the color uniformity of a minimized display screen based on image feature matching according to claim 2, characterized in that, The analysis examines the indirect conduction trajectory of the initial conduction characteristics from directly adjacent light-emitting units to non-directly adjacent light-emitting units, determining the jump nodes, conduction delay, and characteristic attenuation features of the indirect conduction path, including: Real-time chromaticity state change data of directly adjacent light-emitting units after receiving the initial conduction characteristics are collected to determine the conduction diffusion range and stabilization time of the initial conduction characteristics in directly adjacent units. Starting from the stable region of conduction and diffusion within directly adjacent light-emitting units, the transmission trajectory of chromaticity features to non-directly adjacent light-emitting units is traced, and the coordinates of all key points on the trajectory and the corresponding chromaticity feature data are marked. Based on the marked key location points, the jump positions in the process of chromaticity feature transmission from directly adjacent units to non-directly adjacent units are identified. These jump positions serve as jump nodes in the indirect transmission path, and the spatial coordinates and feature transformation states of the jump nodes are recorded. The total time required for the initial conduction characteristics to jump from the main body region of the light-emitting unit to the non-directly adjacent unit is calculated. The time for direct conduction to the adjacent unit is then subtracted to obtain the conduction delay time for indirect conduction. Analyze the changes in chromaticity feature data before and after the jump node to determine the conversion loss and attribute change characteristics of the features during the jump process; Along the indirect transmission trajectory, the system is divided into multiple detection segments at equal intervals, and the chromaticity feature intensity data of the start and end points of each detection segment are extracted. Calculate the attenuation of chromaticity characteristic intensity within each detection segment, and determine the local attenuation coefficient of the detection segment by combining the length of the detection segment and the characteristics of the conductive medium. By integrating the local attenuation coefficients of all detection segments, analyzing their variation characteristics with conduction distance and the distribution of jump nodes, the overall characteristic attenuation characteristics of the indirect conduction path are summarized. The piecewise function is used to fit the overall feature decay feature, generate a mathematical model of the decay feature, and perform an accurate description and adaptation operation of the decay feature. The coordinates of the jump node, the duration of the conduction delay, the feature transition state, and the overall attenuation features after fitting are associated and stored to form complete descriptive data of the indirect conduction trajectory. Based on complete descriptive data, an indirect transmission path model is constructed, which simulates the entire process of chromaticity features being transmitted twice to non-directly adjacent units; The predicted trajectory of the indirect conduction path model is compared with the standard trajectory of different conduction scenarios. If the similarity is lower than a preset threshold, the model parameters are adjusted, and the adaptation operation of the indirect conduction path model and the trajectory simulation standard of different conduction scenarios is performed.

8. The method for correcting the color uniformity of a minimized display screen based on image feature matching according to claim 3, characterized in that, Based on the weighted scene parameters, a correspondence model is established between the scene parameters and the standard conduction threshold, standard conduction attenuation coefficient, and standard superposition method. This correspondence model reflects the influence of different scene parameters on the conduction rules, including: The collected scene parameters under various preset working scenarios are classified and organized to determine the type, scope of influence and quantification method of each scene parameter, forming a scene parameter set; Extract the standard conduction threshold, standard conduction attenuation coefficient and standard superposition method of all light-emitting units in each scenario, and construct a standard rule dataset. The standard rule dataset corresponds one-to-one with the scenario parameter set. The feature mapping algorithm is used to convert each scene parameter in the scene parameter set into a high-dimensional feature vector, which represents the essential attributes and influence weights of the scene parameter. The standard conduction threshold and standard conduction attenuation coefficient in the standard rule dataset are normalized, and the standard superposition method is converted into a standardized encoding vector to ensure that the rule data has a unified computational dimension. A multi-level neural network model is constructed, which includes an input layer, a hidden layer and an output layer. The input layer receives high-dimensional feature vectors of scene parameters, and the output layer outputs corresponding standardized rule data. Configure the hidden layers of the multi-level neural network model to use non-linear activation functions, set their hierarchical connection methods, and perform model feature extraction and mapping capability adaptation operations. The high-dimensional feature vector of scene parameters and the corresponding standardized rule data are used as training samples and input into a multi-level neural network model for training. By adjusting the weight parameters and activation function of the multi-level neural network model, the deviation between the output of the multi-level neural network model and the standard rule data is minimized. During training, cross-validation is used to divide the training set and validation set, and the prediction error of the model on the validation set is evaluated. The multi-level neural network model is used to collect prediction data and generalization data in different scenarios. Based on the feedback data of the validation set, the hyperparameters of the multi-level neural network model are adjusted, and the model prediction output is calibrated. The sensitivity adaptation operation of the multi-level neural network model to changes in scene parameters and the calibration operation of rule prediction data are performed. After training, the test set is used to input the multi-level neural network model, the error between the model output rule data and the actual standard rule data is compared, and the consistency verification operation between the rule data output by the multi-level neural network model and the actual standard rule data is performed. The multi-level neural network model that passed the test was determined as the correspondence model between scene parameters and standard conduction rules. This correspondence model outputs the corresponding standard conduction threshold, standard conduction attenuation coefficient and predicted value of standard superposition method in real time according to the input scene parameters. The output of the correspondence model is compensated and corrected based on historical error data to reduce the difference between the predicted value and the actual standard value.

9. A color uniformity correction device for a minimized display screen based on image feature matching, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the image feature matching-based color uniformity correction method for a minimized display screen as described in any one of claims 1 to 8 by executing the machine-executable instructions.

10. A display screen, characterized in that, The display screen includes: The display panel has multiple minimized light-emitting units arranged on it; The image feature matching-based Miniled display color uniformity correction device is electrically connected to the display panel and configured to perform the image feature matching-based Miniled display color uniformity correction method according to any one of claims 1 to 8.

Images