Miniled display screen brightness regulation method and device based on multi-source environment perception and display screen

Abstract

This invention provides a method, device, and display screen for brightness control of a MiniLED display screen based on multi-source environmental perception. By acquiring dynamic changes in ambient light and content features of the currently displayed image, the method inputs these information into a pre-defined human visual perception model to analyze perception requirements, obtaining a set of perception requirements features matching the current viewing scene. Based on this feature set, a dynamic mapping mechanism is invoked to construct a brightness mapping relationship, generating a dynamic brightness mapping curve reflecting the synergistic effect of ambient light and image content. A pre-constructed partitioned light effect model is then invoked, and inverse solving is performed based on the model and the dynamic brightness mapping curve to generate partitioned calibration driving parameters, which are then sent to the MiniLED backlight module for brightness adjustment. This invention effectively improves the accuracy of brightness control and visual comfort of the MiniLED display screen, while ensuring consistency between the actual hardware output and the target brightness.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus and display screen for adjusting the brightness of a Miniled display screen based on multi-source environmental perception. Background Technology

[0002] With the development of display technology, brightness control of MiniLED displays has become crucial for improving display performance. Existing brightness control methods mostly rely on adjusting a single ambient light parameter or the overall brightness of the image, which makes it difficult to accurately reflect the human eye's perceptual needs for different areas. At the same time, they ignore the luminous efficacy characteristics and crosstalk effects of backlight zones, resulting in insufficient control precision and an inability to balance display performance and visual comfort. Summary of the Invention

[0003] In view of this, the present invention provides a method, apparatus, and display screen for brightness control of a minimized display screen based on multi-source environmental perception. The technical solution of the present invention is implemented as follows:

[0004] On one hand, embodiments of the present invention provide a method for brightness control of a minimized display screen based on multi-source environmental perception. The method includes: acquiring dynamic change information of ambient light in the current environment and content feature information of the currently displayed image on the display screen. The dynamic change information of ambient light includes the distribution of light intensity at different times, and the content feature information includes the brightness and color distribution of each display area in the image. The method further involves inputting the dynamic change information of ambient light and the content feature information into a preset human eye visual perception model to analyze perception requirements, obtaining a set of perception requirement features matching the current viewing scenario. The set of perception requirement features includes indicators of human eye brightness sensitivity to different display areas and visual comfort preference indicators. Finally, the method adjusts the brightness of the minimized display screen based on the perceived brightness and color distribution of each display area. The system invokes a dynamic mapping mechanism to construct a brightness mapping relationship, generating a dynamic brightness mapping curve that reflects the synergistic effect of ambient light and image content. This curve describes the nonlinear transformation relationship from input brightness value to output brightness value. It then invokes a pre-built partitioned light effect model, which includes the luminous efficacy output characteristics of each partition's LEDs under different driving conditions and the crosstalk influence between adjacent partitions. This model is pre-built based on pre-shipment temperature sensitivity test data and crosstalk test data. Finally, it performs a reverse solution based on the partitioned light effect model and the dynamic brightness mapping curve to generate partitioned calibration driving parameters. These parameters are then sent to the drive control unit of the MiniLED backlight module to drive brightness adjustment.

[0005] On the other hand, embodiments of the present invention provide a brightness control device for a minimized display screen, comprising: an information acquisition module, configured to acquire dynamic change information of ambient light in the current environment and content feature information of the currently displayed image on the display screen, wherein the dynamic change information of ambient light includes the distribution of light intensity at different times, and the content feature information includes the brightness distribution and color distribution of each display area in the image; a demand analysis module, configured to input the dynamic change information of ambient light and the content feature information into a preset human eye visual perception model for perception demand analysis, thereby obtaining a perception demand feature set matching the current viewing scenario, wherein the perception demand feature set includes human eye brightness sensitivity index and visual comfort preference index for different display areas; and a relationship construction module, configured to, based on the perception demand feature set... The system invokes a dynamic mapping mechanism to construct a brightness mapping relationship, generating a dynamic brightness mapping curve that reflects the synergistic effect of ambient light and image content. This dynamic brightness mapping curve describes the nonlinear transformation relationship from input brightness value to output brightness value. A model invocation module invokes a pre-built partitioned light effect model, which includes the luminous efficacy output characteristics of each partition's LEDs under different driving conditions and the crosstalk influence between adjacent partitions. This partitioned light effect model is pre-built based on pre-shipment temperature sensitivity test data and crosstalk test data. A drive adjustment module performs inverse solving based on the partitioned light effect model and the dynamic brightness mapping curve to generate partitioned calibration drive parameters. These parameters are then sent to the drive control unit of the MiniLED backlight module to drive brightness adjustment.

[0006] In another aspect, embodiments of the present invention provide a display screen, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the program to implement the steps in the methods described above.

[0007] This invention achieves precise capture of the spatiotemporal changes in ambient light and the spatial details of image content by acquiring dynamic information about ambient light changes in the current environment and content feature information of the currently displayed image. This avoids the one-sided control caused by relying on a single environmental parameter or overall content parameter in traditional control methods. Multi-source dynamic information is input into a preset human visual perception model for perception requirement analysis, resulting in a perception requirement feature set that includes indicators of human eye sensitivity to brightness in different display areas and visual comfort preference indicators. This allows for in-depth modeling of the differentiated perception requirements of the human eye for different areas, breaking through the superficial processing logic of simple ambient light compensation or content brightness adaptation in existing technologies. Based on the perception requirement feature set, a dynamic mapping mechanism is invoked to construct a brightness mapping relationship, generating a dynamic brightness mapping curve that reflects the synergistic effect of ambient light and image content. This curve adapts to the nonlinear characteristics of human vision through a nonlinear transformation relationship, avoiding the limitations of fixed mapping curves or linear mappings that cannot simultaneously consider the synergistic effect of environment and content. The system utilizes a pre-built partitioned light effect model, constructed based on pre-shipment temperature sensitivity and optical crosstalk test data. This model accurately characterizes the light effect output characteristics of each partition's LEDs under different driving conditions and the impact of optical crosstalk between adjacent partitions, resolving the issue of actual output deviation caused by neglecting non-ideal hardware characteristics in traditional control methods. By reverse-engineering the partitioned light effect model and dynamic brightness mapping curve, partitioned calibration driving parameters are generated. This ensures that the calculated driving parameters not only meet human visual perception requirements but also adapt to the hardware's temperature sensitivity and crosstalk characteristics, effectively improving the accuracy of brightness control and human visual comfort in minimized displays, while simultaneously guaranteeing consistency between the actual hardware output and the target brightness. Attached Figure Description

[0008] Figure 1 This is a schematic diagram illustrating the implementation process of a brightness control method for a Miniled display screen based on multi-source environmental perception, provided in an embodiment of the present invention.

[0009] Figure 2 This is a schematic diagram of the composition structure of a Miniled display screen brightness control device provided in an embodiment of the present invention.

[0010] Figure 3 This is a schematic diagram of the hardware entity of a display screen provided in an embodiment of the present invention. Detailed Implementation

[0011] This invention provides a method for brightness control of a minimized display screen based on multi-source environmental perception. This method can be implemented by the display screen's processor executing a program stored in its memory. The display screen can refer to high-end home television displays, professional gaming displays, large commercial advertising displays, or other displays integrating multiple sensors and high computing power.

[0012] Figure 1 This is a schematic diagram illustrating the implementation process of a brightness control method for a minimized display screen based on multi-source environmental perception, as provided in an embodiment of the present invention. Figure 1 As shown, the method includes:

[0013] Step S100: Obtain the ambient light dynamic change information under the current environment and the content feature information of the image currently displayed on the screen. The ambient light dynamic change information includes the light intensity distribution at different times, and the content feature information includes the brightness distribution and color distribution of each display area in the image.

[0014] Ambient light dynamic change information refers to the distribution of light intensity in the current environment at different times over time. This information reflects real-time changes in ambient light. For example, during the day, the intensity and direction of ambient light constantly change as the sun moves; at night, the light intensity may fluctuate due to the influence of artificial light. Ambient light dynamic change information can be obtained using light sensors, such as by evenly distributing multiple light sensors around a display screen and recording the light intensity value at regular intervals, thus obtaining the light intensity distribution at different times.

[0015] The content feature information of the image currently displayed on the screen includes the brightness and color distribution of each display area. Brightness distribution reflects the brightness of different areas in the image; for example, in a landscape photograph, the sky may be brighter, while the shadows of trees may be darker. Color distribution reflects the variety and proportion of colors in different areas of the image; for example, in a painting, there may be a large number of red and yellow areas. Content feature information can be obtained through image analysis techniques. The image on the screen is segmented into multiple display areas, and then the brightness and color of each area are statistically analyzed. For example, image processing algorithms can be used to convert the image into a digital matrix, and the brightness and color information is determined by calculating the value of each element in the matrix.

[0016] Step S200: Input the ambient light dynamic change information and content feature information into the preset human eye visual perception model to analyze the perception requirements and obtain a perception requirement feature set that matches the current viewing scene. The perception requirement feature set includes the human eye's brightness sensitivity index and visual comfort preference index for different display areas.

[0017] The human eye visual perception model is a complex model used to simulate the human eye's perception process of different ambient light and image content. It comprehensively considers the physiological characteristics and visual psychological factors of the human eye, and can parse out the set of perceptual requirement features that match the current viewing scene based on the dynamic changes of the input ambient light and the content feature information of the image currently displayed on the screen.

[0018] For example, the model can employ a multi-layer neural network architecture, mainly composed of an input layer, hidden layers, and an output layer. The input layer receives ambient light dynamic change information and content feature information presented in vector or matrix form. The ambient light dynamic change information includes the light intensity distribution at different times, which can be represented as a light intensity vector changing over time. The content feature information includes the brightness and color distribution of each display area in the image, which can be represented as a matrix containing the brightness and color information of each display area. The hidden layer, as the core of the model, consists of multiple neuron layers, each containing multiple neurons. For example, it includes a temporal environmental feature processing sublayer, a regional content feature processing sublayer, a cross-sensory association processing sublayer, a threshold adjustment processing sublayer, a hierarchical parsing processing sublayer, and so on.

[0019] The temporal environmental feature processing sublayer can, for example, employ a convolutional layer structure. A convolutional layer consists of multiple convolutional kernels that perform sliding convolution operations on the input ambient light dynamics information (represented as a time-varying light intensity vector). This operation automatically extracts local features from the input information; for ambient light dynamics information, the convolutional kernels can capture the fluctuation patterns of light intensity over different time periods. For example, some convolutional kernels may be specifically designed to detect rapid increases or decreases in light intensity, while others are used to identify relatively gentle fluctuation phases. After the convolution operation, nonlinear activation, such as using the ReLU function, is performed to introduce nonlinearity and enhance the model's expressive power. By setting an appropriate number and size of convolutional kernels, this sublayer can accurately divide the light intensity change cycle into a first fluctuation segment and a second fluctuation segment, and extract the fluctuation amplitude and duration features of each segment. The fluctuation amplitude feature can be determined by the difference between the maximum and minimum values ​​in the convolutional kernel output, while the duration feature can be calculated based on the response range of the convolutional kernel in the time dimension. Finally, these features are combined to generate a temporal environmental feature sequence.

[0020] The region content feature processing sublayer can be constructed as an autoencoder structure, consisting of an encoder and a decoder. The encoder compresses the input matrix containing the brightness and color distributions of each display region in the image, mapping high-dimensional image feature information to a low-dimensional hidden representation space. During this process, the autoencoder learns the essential features of each region in the image, enabling it to assign visual weights to the display regions. For example, important target regions in the image are given higher weights in the hidden representation. The decoder then remaps the low-dimensional hidden representation back to the high-dimensional space, reconstructing a result similar to the input image. By comparing the differences between the input and reconstructed images, the autoencoder can continuously optimize its parameters. In this process, the autoencoder can identify first-class and second-class regions in the image, based on the feature distribution in the hidden representation. Regions with complex brightness and color variations are likely classified as first-class regions, while relatively simple regions are classified as second-class regions. Simultaneously, this sublayer calculates the brightness distribution density and color transition smoothness of each region. Brightness distribution density can be measured by calculating the variance of brightness values ​​within a region of the reconstructed image; a smaller variance indicates a more uniform brightness distribution and higher density. Color transition smoothness can be determined by analyzing the gradient of color values ​​between adjacent pixels in the reconstructed image; a smaller gradient indicates a smoother color transition. Finally, these calculation results are combined to generate a region content feature matrix.

[0021] The cross-sensing correlation processing sublayer can employ a graph neural network (GNN) structure, treating the temporal environmental feature sequence and regional content feature matrix as information from the nodes and edges of a graph. Nodes in the graph can represent different fluctuation segments or display areas, while edges represent the relationships between them. The GNN transmits information between the nodes of the graph through a message-passing mechanism, continuously updating the feature representations of the nodes. In this process, the GNN can calculate the response correlation degree between the first fluctuation segment and the first type of region, and the adaptive correlation degree between the second fluctuation segment and the second type of region. The response correlation degree can be obtained by calculating the similarity of features between nodes, for example, using cosine similarity to measure the similarity between fluctuation segment features and region features; the adaptive correlation degree can be determined by analyzing the dynamic changes of nodes during the message-passing process. Finally, this correlation information is filled into a pre-defined matrix framework to generate a cross-sensing correlation matrix.

[0022] The threshold adjustment sublayer can be designed as a Long Short-Term Memory (LSTM) network structure. LSTM is particularly suitable for processing sequential data; the input cross-sensory association matrix and region content feature matrix can be viewed as sequential information with temporal or logical order. LSTM includes input gates, forget gates, and output gates, and these gating mechanisms can effectively handle long-term dependencies in the sequence. When processing the cross-sensory association matrix, LSTM performs threshold filtering based on the order of matrix elements and contextual information, comparing matrix elements with a preset association strength threshold and filtering out elements greater than the threshold. Next, based on the row and column indices corresponding to the filtered elements, the corresponding environmental fluctuation segment features and region content features are extracted from the temporal environmental feature sequence and region content feature matrix, and combined into multiple feature query pairs. Based on these feature query pairs, LSTM queries a preset visual perception threshold library to obtain the corresponding standard brightness perception threshold and standard color perception threshold. Then, these thresholds are adjusted according to the region content feature matrix, which can be achieved through a linear combination between the feature vector output by LSTM and the region content features. Finally, the adjusted thresholds are arranged in regional order to generate a dynamic perception threshold sequence.

[0023] The hierarchical parsing sublayer can employ a structure combining an attention mechanism and a fully connected layer. The input dynamic perception threshold sequence first undergoes processing through the attention mechanism. This mechanism automatically assigns weights to thresholds in different regions, highlighting threshold information in important areas. Attention weights are obtained by calculating the similarity between each element in the threshold sequence and a learnable query vector. Then, the attention weights are weighted and summed with the threshold sequence to obtain the attention-enhanced threshold representation. This representation is then input into the fully connected layer. The fully connected layer performs linear transformations and non-linear activation processing on the input, first parsing the brightness sensitivity component of the first type of region, and then parsing the visual comfort component of the second type of region. The brightness sensitivity component can be determined by analyzing the threshold changes in the first type of region after attention enhancement and the sensitivity characteristics of the human eye to brightness changes. The visual comfort component can be obtained by comprehensively considering the relevant features of the second type of region and the visual psychological needs of the human eye. When calculating the changing trends of each component in the temporal environmental feature sequence, the fully connected layer performs correlation analysis on the temporal environmental feature sequence and each component, determining the changing trends by calculating statistical measures such as their covariance. Finally, these trends are represented as vectors to generate trend feature vectors.

[0024] In one implementation, step S200 may specifically include the following steps S210 to S260:

[0025] Step S210: Perform time-domain fluctuation sampling on the dynamic change information of ambient light, divide it into the first fluctuation segment and the second fluctuation segment according to the light intensity change cycle, extract the fluctuation amplitude features and duration features of each segment, and generate a time-domain environmental feature sequence.

[0026] Temporal fluctuation sampling involves sampling the dynamic changes in ambient light over time to obtain information on how light intensity changes over time. By using temporal fluctuation sampling, continuous ambient light variation information can be discretized, dividing it into first and second fluctuation segments based on the period of light intensity change. This segmentation is based on the pattern of light intensity variation. For example, light intensity may fluctuate multiple times throughout the day; dividing these fluctuations according to their period of change yields different fluctuation segments.

[0027] The amplitude of light fluctuation is the maximum change in light intensity within each fluctuation segment, reflecting the drastic change in ambient light within that segment. The duration of light fluctuation is the duration of each segment, reflecting the stability of ambient light within that segment. Both amplitude and duration characteristics can be extracted by analyzing and calculating the sampled light intensity data. For example, for each fluctuation segment, finding the maximum and minimum light intensity values ​​and calculating their difference yields the amplitude characteristic; recording the start and end times of the fluctuation segment and calculating their difference yields the duration characteristic.

[0028] A temporal environmental feature sequence is a sequence that includes the amplitude and duration characteristics of each fluctuation segment, comprehensively reflecting the changing characteristics of ambient light over different time periods. By generating a temporal environmental feature sequence, the dynamic changes in ambient light can be converted into a feature representation that is easy to process and analyze.

[0029] Step S220: Perform regional visual weight labeling on the image display area in the content feature information, identify the first type of region and the second type of region in the image, calculate the brightness distribution density and color transition smoothness of each region, and generate the regional content feature matrix.

[0030] Region visual weighting assigns a weight value to each display area based on the human eye's attention to and importance of different areas of an image. For example, in an image, the central area is usually the focus of human attention, so it can be assigned a higher weight value; while the edge areas have relatively lower attention and can be assigned lower weight values. Identifying first-class and second-class regions in an image can be done based on the image's content and structure. For example, in a photograph of a person, the area containing the person can be classified as a first-class region, while the background area can be classified as a second-class region.

[0031] Brightness distribution density refers to the distribution of brightness values ​​within each region, reflecting the uniformity of brightness within that region. Color transition smoothness, on the other hand, describes the smoothness of color changes within each region, reflecting the continuity of color within that region. Brightness distribution density and color transition smoothness can be calculated through statistical analysis of the image's brightness and color data. For example, for each region, the standard deviation of its brightness values ​​can be calculated; a smaller standard deviation indicates a more uniform brightness distribution and a higher brightness distribution density. Similarly, color gradient algorithms can be used to calculate the smoothness of color transitions; a smaller gradient value indicates a smoother color transition.

[0032] The regional content feature matrix is ​​a matrix that includes the brightness distribution density and color transition smoothness of each region, and it can comprehensively reflect the content features of each display area in the image.

[0033] Step S230: Cross-dimensional correlation between the temporal environmental feature sequence and the regional content feature matrix, calculate the response correlation degree between the first fluctuation segment and the first type of region, and the adaptation correlation degree between the second fluctuation segment and the second type of region, and generate a cross-sensory correlation matrix.

[0034] Cross-dimensional correlation involves analyzing the correlation between temporal environmental feature sequences and regional content feature matrices across different dimensions to identify their intrinsic relationships. The response correlation between the first fluctuation segment and the first type of region reflects the degree to which changes in ambient light within the first fluctuation segment affect the first type of region; for example, a sudden increase in light intensity may cause corresponding changes in the brightness and color of the first type of region. The adaptation correlation between the second fluctuation segment and the second type of region reflects the adaptability of the second type of region to changes in ambient light within the second fluctuation segment; for example, with slow changes in light intensity, the brightness and color of the second type of region may gradually adapt to this change.

[0035] In one implementation, step S230 may specifically include the following steps S231 to S236:

[0036] Step S231: Expand the time-domain environmental feature sequence in terms of dimensions, normalize the fluctuation amplitude and duration features of the first fluctuation segment, and convert them into a dimensionless two-dimensional fluctuation feature matrix. Similarly, convert the features of the second fluctuation segment into a two-dimensional stable feature matrix.

[0037] Dimensional expansion further decomposes and expands the features of each fluctuation segment in the time-domain environmental feature sequence to facilitate subsequent processing and analysis. Normalization unifies feature data of different scales and ranges, giving them the same dimensions and range. For example, the numerical ranges of fluctuation amplitude and duration features may differ; normalization converts them into dimensionless values ​​for easier comparison and calculation. After normalizing the fluctuation amplitude and duration features of the first fluctuation segment, a dimensionless two-dimensional fluctuation feature matrix is ​​generated. The rows and columns of the matrix represent different features and samples, respectively, and each element represents the normalized value of the corresponding feature on the corresponding sample. The features of the second fluctuation segment are similarly converted into a two-dimensional stable feature matrix.

[0038] Step S232: Standardize the region size and normalize the features of the region content feature matrix to unify the matrix dimensions and numerical range of each display region, so that the number of rows, columns and numerical scale of the feature matrix of the first type region and the second type region are consistent, and generate a standardized region feature matrix.

[0039] Region size standardization unifies the dimensions of each display region in the region content feature matrix, ensuring they have the same matrix dimensions. For example, different display regions may differ in size; region size standardization converts them into matrices of the same size. Feature normalization unifies the numerical ranges of each feature in the region content feature matrix, ensuring they have the same dimensions and ranges. For example, brightness distribution density and color transition smoothness may have different numerical ranges; feature normalization converts them into dimensionless values, facilitating comparison and calculation. Generating a standardized region feature matrix involves applying region size standardization and feature normalization to the region content feature matrix, ensuring that the number of rows, columns, and numerical scales of the feature matrices for both the first and second type of regions remain consistent.

[0040] Step S233: Set a fixed calculation window on the standardized regional feature matrix. The window size matches the dimension of the two-dimensional fluctuation feature matrix. Slide the window in row priority order and query the association weight values ​​of the normalized regional features and the corresponding elements of the two-dimensional fluctuation feature matrix based on the predefined association rule mapping table.

[0041] A fixed computation window is set on the standardized regional feature matrix, with the window size matching the dimension of the two-dimensional fluctuation feature matrix. A row-major sliding window starts from the first row of the standardized regional feature matrix and slides row by row until the entire matrix has been traversed. A predefined association rule mapping table contains the association weight values ​​between the normalized regional features and their corresponding elements in the two-dimensional fluctuation feature matrix; this table is derived from extensive experimental data and empirical observations. During the sliding window process, for each normalized regional feature and its corresponding element in the two-dimensional fluctuation feature matrix within each window, their association weight values ​​are queried according to the predefined association rule mapping table. By setting a fixed computation window and querying association weight values, association analysis can be performed on the standardized regional feature matrix and the two-dimensional fluctuation feature matrix to identify their intrinsic relationships.

[0042] Step S234: Dynamically weight and sum the associated weight values ​​of all windows corresponding to the first fluctuation segment and the first type of region to generate a weighted response correlation degree. The weight value is determined according to the weight level in the region visual weight label. The higher the weight level, the larger the weight value.

[0043] Dynamic weighted summation involves assigning a weight coefficient to each associated weight value based on the weight level in the visual weight label of the region when summing all window association weight values ​​corresponding to the first fluctuation segment and the first type of region. A higher weight level results in a larger weight coefficient, indicating greater importance of the region in calculating the weighted response association degree. The weighted response association degree is generated by using the result of the dynamic weighted summation as the response association degree between the first fluctuation segment and the first type of region.

[0044] Step S235: Accumulate the associated weight values ​​of all windows corresponding to the second fluctuation segment and the second type of region, and add the associated weight values ​​in order of spatial location of the region in the image to generate the cumulative adaptive correlation degree.

[0045] Cumulative calculation involves adding up the association weights of all windows corresponding to the second wave segment and the second type of region according to the spatial position of the region in the image. For example, starting from the top left corner of the image, the association weights are added row by row and column by column. The cumulative adaptive correlation score is generated by using the result of the cumulative calculation as the adaptive correlation score between the second wave segment and the second type of region.

[0046] Step S236: Fill the weighted response correlation degree and cumulative adaptive correlation degree into the preset matrix frame according to the row and column correspondence. The first fluctuation segment corresponds to the matrix row, the first type of region corresponds to the matrix column, and the second fluctuation segment and the second type of region are similarly generated to generate the cross-sensing correlation matrix.

[0047] The preset matrix framework is a predefined matrix structure, with the number of rows and columns corresponding to the fluctuation segments in the time-domain environmental feature sequence and the display areas in the regional content feature matrix, respectively. The weighted response correlation and cumulative adaptive correlation are filled into the preset matrix framework according to their row-column correspondence. Specifically, the weighted response correlation between the first fluctuation segment and the first type of region is filled into the row corresponding to the first fluctuation segment and the column corresponding to the first type of region in the matrix; the cumulative adaptive correlation between the second fluctuation segment and the second type of region is filled into the row corresponding to the second fluctuation segment and the column corresponding to the second type of region in the matrix.

[0048] Step S240: Based on the cross-sensing association matrix, query the preset visual perception threshold library to obtain the brightness perception threshold and color perception threshold of each region under the current ambient light fluctuation. Adjust the dynamic range of each threshold according to the region content feature matrix to generate a dynamic perception threshold sequence.

[0049] The preset visual perception threshold database is a collection of brightness and color perception thresholds for various regions under different ambient light conditions. It was established based on extensive experimental data and research findings. Querying the preset visual perception threshold database involves using the correlation information in the cross-perception association matrix to find the brightness and color perception thresholds for each region under the current ambient light fluctuations.

[0050] Adjusting the dynamic range of each threshold based on the regional content feature matrix involves adjusting the luminance perception threshold and color perception threshold obtained from the query based on the luminance distribution density and color transition smoothness of each region. For example, if a region has a high luminance distribution density, it indicates that the brightness in that region is relatively uniform, and the dynamic range of the luminance perception threshold can be appropriately reduced; if a region has a high color transition smoothness, it indicates that the color changes in that region are relatively continuous, and the dynamic range of the color perception threshold can be appropriately reduced.

[0051] In one implementation, step S240 may specifically include the following steps S241 to S247:

[0052] Step S241: Perform threshold filtering on the cross-sensing correlation matrix, compare the matrix elements with the preset correlation strength threshold, filter out elements that are greater than the correlation strength threshold, record the row index and column index corresponding to each filtered element, and generate a list of strongly correlated elements, wherein the row index is associated with the corresponding fluctuation segment in the time domain environmental feature sequence, and the column index is associated with the corresponding display area in the regional content feature matrix.

[0053] Threshold filtering compares each element in the cross-sense correlation matrix with a preset correlation strength threshold, selecting elements that are greater than the threshold. The preset correlation strength threshold is a pre-defined value used to determine whether the correlation between elements is strong enough. For example, if the correlation strength threshold is 0.5, only elements in the cross-sense correlation matrix with a value greater than 0.5 will be filtered out. The row and column indices corresponding to each filtered element are recorded to generate a list of strongly correlated elements. The row index is associated with the corresponding fluctuation segment in the time-domain environmental feature sequence, and the column index is associated with the corresponding display area in the regional content feature matrix.

[0054] Step S242: Based on the row and column indices in the strongly correlated element list, extract the corresponding environmental fluctuation segment features and regional content features from the temporal environmental feature sequence and regional content feature matrix, and combine them into multiple feature query pairs.

[0055] Based on the row and column indices in the strongly correlated element list, the corresponding environmental fluctuation segment features are extracted from the temporal environmental feature sequence, and the corresponding regional content features are extracted from the regional content feature matrix. For example, if an element in the strongly correlated element list has a row index of 1 and a column index of 2, then the features of the first fluctuation segment can be extracted from the temporal environmental feature sequence, and the features of the second type of region can be extracted from the regional content feature matrix.

[0056] The extracted environmental fluctuation segment features and regional content features are combined into multiple feature query pairs. Each feature query pair contains one environmental fluctuation segment feature and one regional content feature, which are used for subsequent query operations. By combining them into multiple feature query pairs, the correlation information between ambient light fluctuations and image display areas can be transformed into a form that is easy to query.

[0057] Step S243: Based on each set of feature query pairs, query the preset visual perception threshold library, obtain the corresponding standard brightness perception threshold and standard color perception threshold, and generate multiple sets of basic thresholds.

[0058] Based on each feature query pair, a preset visual perception threshold library is queried. This involves finding the corresponding standard brightness perception threshold and standard color perception threshold from the visual perception threshold library based on the environmental fluctuation segment features and regional content features in each feature query pair. The standard brightness perception threshold and standard color perception threshold are the human eye's perception thresholds for brightness and color under specific ambient light conditions and regional content features.

[0059] Generating multiple sets of basic thresholds involves combining the obtained standard brightness perception thresholds and standard color perception thresholds into multiple sets of basic thresholds. Each set of basic thresholds contains one standard brightness perception threshold and one standard color perception threshold, which are used for subsequent fusion and adjustment operations.

[0060] Step S244: Merge multiple sets of basic thresholds. For each display area, take a weighted average of all the basic thresholds associated with it based on the correlation values ​​of the corresponding strongly correlated elements to generate the initial fused brightness threshold and the initial fused color threshold for that area.

[0061] The fusion of multiple sets of basic thresholds involves comprehensively processing all the basic thresholds associated with each display area. Specifically, a weighted average is calculated based on the correlation values ​​of corresponding strongly correlated elements. The larger the correlation value, the stronger the correlation with that basic threshold, and the greater its weight in the weighted average. The initial fused brightness threshold and initial fused color threshold for that area are generated by using the weighted average brightness perception threshold and color perception threshold as the initial fused brightness threshold and initial fused color threshold for that area, respectively.

[0062] Step S245: Input the brightness distribution density and color transition smoothness of the corresponding region in the region content feature matrix into the predefined brightness adjustment mapping function and color adjustment mapping function, respectively. Through the brightness adjustment mapping function, the brightness distribution density is mapped to a first dimensionless adjustment factor. The brightness adjustment mapping function is configured such that the higher the brightness distribution density, the larger the output first dimensionless adjustment factor. Through the color adjustment mapping function, the color transition smoothness is mapped to a second dimensionless adjustment factor. The color adjustment mapping function is configured such that the higher the color transition smoothness, the smaller the output second dimensionless adjustment factor. The first dimensionless adjustment factor is used as the brightness threshold adjustment coefficient, and the second dimensionless adjustment factor is used as the color threshold adjustment coefficient.

[0063] The predefined brightness and color adjustment mapping functions are predefined functions used to map brightness distribution density and color transition smoothness into dimensionless adjustment factors. The brightness adjustment mapping function is configured such that the higher the brightness distribution density, the larger the output first dimensionless adjustment factor. This is because a higher brightness distribution density indicates a more uniform brightness in the area, which may reduce the human eye's sensitivity to brightness, thus requiring a larger adjustment factor to adjust the brightness perception threshold. The color adjustment mapping function is configured such that the higher the color transition smoothness, the smaller the output second dimensionless adjustment factor. This is because a higher color transition smoothness indicates a more continuous color change in the area, which may increase the human eye's sensitivity to color, thus requiring a smaller adjustment factor to adjust the color perception threshold. The brightness adjustment mapping function can be a piecewise defined function, and the color adjustment mapping function can be an exponential function; the specific limitations are not specified.

[0064] Step S246: By adjusting the brightness threshold and color threshold, the dynamic range of the initial fused brightness threshold and the initial fused color threshold are scaled respectively to generate the final brightness perception threshold and the final color perception threshold for each region.

[0065] The dynamic range of the initial fused brightness threshold and the initial fused color threshold are scaled using brightness threshold adjustment coefficients and color threshold adjustment coefficients, respectively. For example, multiplying the initial fused brightness threshold by the brightness threshold adjustment coefficient yields the final brightness perception threshold; multiplying the initial fused color threshold by the color threshold adjustment coefficient yields the final color perception threshold.

[0066] The final luminance and color perception thresholds for each region are generated by using scaled luminance and color perception thresholds as the final perception thresholds for each region. By scaling the dynamic range of the initial fusion thresholds, the perception thresholds for each region can be adjusted more accurately based on the region's content characteristics and ambient light fluctuations.

[0067] Step S247: Arrange the final brightness perception threshold and final color perception threshold of all regions in the order of regions to generate a dynamic perception threshold sequence.

[0068] The final brightness perception threshold and final color perception threshold of all regions are arranged in region order, that is, the final brightness perception threshold and final color perception threshold of each region are arranged sequentially into a sequence. The region order can be determined based on the arrangement order of each region in the region content feature matrix.

[0069] Step S250: Perform hierarchical analysis on the dynamic perception threshold sequence. First, analyze the brightness sensitivity component of the first type of region, then analyze the visual comfort component of the second type of region, calculate the changing trend of each component in the temporal environmental feature sequence, and generate a trend feature vector.

[0070] Hierarchical analysis involves performing layered analysis on the dynamic perception threshold sequence, separately analyzing different components of the first and second category regions. First, the luminance sensitivity component of the first category region is analyzed, reflecting the human eye's sensitivity to changes in brightness within this region. For example, in an image, the first category region is often the focus of human attention, thus exhibiting high sensitivity to changes in its brightness. Next, the visual comfort component of the second category region is analyzed, reflecting the level of comfort experienced by the human eye when viewing this region. For example, the second category region might be the background area, where the human eye has relatively high requirements for visual comfort. Calculating the changing trends of each component within the temporal environmental feature sequence analyzes how the luminance sensitivity and visual comfort components change over time. For instance, with fluctuations in ambient light, the luminance sensitivity and visual comfort components may change accordingly.

[0071] Step S260: Dynamically assemble the brightness sensitivity component, visual comfort component, and trend feature vector, sort them by region visual weight to construct a perceptual demand feature set, so that the sensitivity feature weight of the first type of region is higher than that of the second type of region, and generate a perceptual demand feature set that matches the current viewing scene.

[0072] Dynamic assembly involves combining and integrating brightness sensitivity components, visual comfort components, and trend feature vectors. For example, they can be arranged in a certain order to form a new feature representation. Constructing a perceptual requirement feature set by sorting by region visual weight involves assigning a weight value to each feature in each region based on its visual weight label, and then sorting these features from highest to lowest weight. The sensitivity feature weights of the first type of region are higher than those of the second type of region because the first type of region is usually the focus of human eye attention, and therefore has higher requirements for brightness sensitivity and visual comfort. Generating a perceptual requirement feature set that matches the current viewing scene involves using the dynamically assembled and sorted feature set as the perceptual requirement feature set that matches the current viewing scene.

[0073] Step S300: Based on the perceptual requirement feature set, the dynamic mapping mechanism is invoked to construct the brightness mapping relationship, generating a dynamic brightness mapping curve that reflects the synergistic effect between ambient light and image content. The dynamic brightness mapping curve is used to describe the nonlinear transformation relationship from input brightness value to output brightness value.

[0074] Dynamic mapping is a mechanism that can dynamically adjust the mapping relationship based on different input conditions. In this method, the dynamic mapping mechanism constructs a brightness mapping relationship based on the perceptual requirement feature set. The brightness mapping relationship is the correspondence between the input brightness value and the output brightness value, reflecting the brightness value that the display screen should output under different ambient light and image content conditions.

[0075] Generating a dynamic brightness mapping curve that reflects the interaction between ambient light and image content represents the brightness mapping relationship as a curve. This curve can intuitively show the non-linear transformation relationship from input brightness value to output brightness value. For example, under different ambient light and image content conditions, the same input brightness value may correspond to different output brightness values. The dynamic brightness mapping curve is used to describe this non-linear transformation relationship.

[0076] In one implementation, step S300 may specifically include the following steps S310 to S360:

[0077] Step S310: The perceptual demand feature set is layered according to the spatial location of the display area, divided into a central area feature layer, a transition area feature layer and an edge area feature layer. Each layer contains the brightness sensitivity index and visual comfort preference index of the corresponding area.

[0078] Layering the perceptual demand feature set according to the spatial location of the display area involves dividing the perceptual demand feature set into different levels based on the spatial location of each display area on the screen. The central area feature layer includes brightness sensitivity and visual comfort preference indicators for the central area of ​​the display screen. The central area is usually the focus of human eye attention, and its brightness and visual comfort requirements are relatively high. The transition area feature layer includes brightness sensitivity and visual comfort preference indicators for the transition areas of the display screen. The transition areas are the parts between the central and edge areas, and their brightness and visual comfort requirements are between those of the central and edge areas. The edge area feature layer includes brightness sensitivity and visual comfort preference indicators for the edge areas of the display screen. The edge areas receive relatively less attention, and their brightness and visual comfort requirements are relatively lower.

[0079] Step S320: Perform temporal sampling on the central region feature layer, the transition region feature layer, and the edge region feature layer respectively, extract the feature values ​​of each layer at different times according to a preset time interval, and generate the temporal feature sequence of each layer.

[0080] Temporal sampling involves sampling features at each layer along the time dimension to capture how these features change over time. Feature values ​​for each layer are extracted at preset time intervals, which can be set according to specific needs, such as extracting feature values ​​every second. Generating a temporal feature sequence for each layer involves arranging the extracted feature values ​​at different times into a chronological sequence, where each element represents the feature value of the corresponding layer at that specific time point. By generating temporal feature sequences for each layer, the temporal variation information of each layer's features can be converted into a sequence format that is easy to process and analyze.

[0081] Step S330: Calculate the correlation strength between the feature layer of the central region and the feature layer of the transition region, and the correlation strength between the feature layer of the transition region and the feature layer of the edge region. The correlation strength is represented by the ratio of the sum of the product of the feature values ​​to the sum of the squares of the feature values, and the interlayer correlation strength value is generated.

[0082] In one implementation, step S330 may specifically include the following steps S331 to S336:

[0083] Step S331: Align the temporal feature sequences of the central region feature layer, the transition region feature layer, and the edge region feature layer with lengths, and truncate the overlapping timestamps in each sequence so that each sequence contains the same number of feature value samples.

[0084] Length alignment unifies the lengths of the temporal feature sequences from the central region feature layer, transition region feature layer, and edge region feature layer, ensuring they contain the same number of feature value samples. Truncation involves identifying and retaining the portions of timestamp overlap between sequences. For example, if the temporal feature sequence of the central region feature layer ranges from 0 to 10 seconds, and the temporal feature sequence of the transition region feature layer ranges from 2 to 12 seconds, then the portion with overlapping timestamps, i.e., from 2 to 10 seconds, is truncated.

[0085] Step S332: Convert the brightness sensitivity index and visual comfort preference index in the time domain feature sequence of each layer into standardized feature values. By subtracting the mean from the feature value and then dividing by the standard deviation, the feature values ​​are made to be in the same numerical range.

[0086] Converting the brightness sensitivity index and visual comfort preference index in each layer of the time-domain feature sequence into standardized feature values ​​is to eliminate differences in the dimensions and numerical ranges between different feature values. Standardization is performed by subtracting the mean from the feature value and then dividing by the standard deviation. For example, for a certain feature value, first calculate the mean and standard deviation of the sequence containing that feature value, then subtract the mean from the feature value and divide by the standard deviation to obtain the standardized feature value.

[0087] Ensuring that all feature values ​​fall within the same numerical range facilitates subsequent calculations and comparisons. Through standardization, the brightness sensitivity and visual comfort preference indices of different layers can be converted into standardized feature values ​​with the same numerical range.

[0088] Step S333: Construct the covariance matrix between the central region and the transition region, and the covariance matrix between the transition region and the edge region. The matrix elements are the covariance of the corresponding feature value sequences. The rows represent the features of the central region or the transition region, and the columns represent the features of the transition region or the edge region.

[0089] Constructing covariance matrices between the central and transitional regions, and between the transitional and edge regions, is to analyze the correlation of features between different layers. The elements of the covariance matrix are the covariances of the corresponding eigenvalue sequences; the covariance reflects the correlation between two variables. For example, for the central and transitional feature layers, the covariances between their brightness sensitivity index and visual comfort preference index are calculated, and these covariance values ​​are used as elements of the covariance matrix.

[0090] The rows of the matrix represent features of the central region or the transition region, and the columns represent features of the transition region or the edge region. By constructing a covariance matrix, the correlation between features of different layers can be represented in matrix form.

[0091] Step S334: Calculate the trace of the covariance matrix as the numerator for calculating the correlation strength. The larger the trace value, the larger the sum of the covariances of the two layers of features.

[0092] The trace of the covariance matrix is ​​calculated by summing the diagonal elements of the matrix. The trace, being the sum of the diagonal elements, reflects the total sum of the main diagonal elements. As a numerator in the correlation strength calculation, a larger trace value indicates a larger sum of covariances between the two layers of features, suggesting a stronger correlation between the features in the two layers.

[0093] Step S335: Calculate the sum of squares of the eigenvalues ​​of the central region feature layer and the transition region feature layer. Add the two together and use them as the denominator for calculating the correlation strength. The larger the sum of squares, the greater the total energy of the two feature layers.

[0094] Calculating the sum of squares of the eigenvalues ​​of the central region feature layer and the transition region feature layer involves squaring the eigenvalues ​​of each layer separately and then summing them. This sum is used as the denominator in the correlation strength calculation; a larger sum indicates a greater total energy between the two feature layers. For example, for the brightness sensitivity index and visual comfort preference index of the central region feature layer, their eigenvalues ​​are squared and summed; similarly, for the brightness sensitivity index and visual comfort preference index of the transition region feature layer, their eigenvalues ​​are squared and summed. The two sums are then added together and used as the denominator in the correlation strength calculation.

[0095] Step S336: Divide the numerator by the denominator to obtain the interlayer correlation strength value, and generate the central transition correlation strength value and the transition edge correlation strength value.

[0096] The inter-layer correlation strength value is obtained by dividing the numerator by the denominator, which is equivalent to dividing the trace of the covariance matrix by the sum of squared eigenvalues. This value ranges from 0 to 1, with a value closer to 1 indicating a higher degree of correlation between the two layers. For example, dividing the numerator of the correlation strength calculation between the central region feature layer and the transition region feature layer by the denominator yields the central-transition correlation strength value; dividing the numerator of the correlation strength calculation between the transition region feature layer and the edge region feature layer by the denominator yields the transition-edge correlation strength value. These generated correlation strength values ​​are then used as the correlation strength values ​​between the central region feature layer and the transition region feature layer, and between the transition region feature layer and the edge region feature layer, respectively.

[0097] Step S340: Generate the brightness mapping base curve of the corresponding region based on the temporal feature sequence of each layer. The base curve of the central region is dominated by the brightness sensitivity index, the base curve of the edge region is dominated by the visual comfort preference index, and the base curve of the transition region combines the two.

[0098] The generation of luminance mapping base curves for corresponding regions based on the temporal feature sequences of each layer involves using luminance sensitivity indices and visual comfort preference indices from the temporal feature sequences of each layer to generate the corresponding luminance mapping curves. The base curve for the central region is primarily driven by the luminance sensitivity index, because the central region is usually the focus of human eye attention and is highly sensitive to changes in its luminance. Therefore, when generating the base curve for the central region, the luminance sensitivity index is the main consideration.

[0099] The baseline curve for the edge region is primarily driven by visual comfort preference indices. Because edge regions receive relatively less attention, their visual comfort requirements are relatively higher. Therefore, when generating the baseline curve for the edge region, visual comfort preference indices are the main consideration. The baseline curve for the transition region combines both factors. Since the transition region is the part between the central and edge regions, its brightness and visual comfort requirements are between those of the central and edge regions. Therefore, when generating the baseline curve for the transition region, both brightness sensitivity indices and visual comfort preference indices need to be considered.

[0100] Step S350: Perform interlayer fusion of the brightness mapping base curves based on the interlayer correlation strength value. The higher the correlation strength value, the greater the weight of the base curve of the adjacent layer. The three base curves are fused into a preliminary brightness mapping curve by weighted averaging.

[0101] Inter-layer fusion of the luminance mapping base curves based on inter-layer correlation strength values ​​involves fusing the base curves of the central region, transition region, and edge region. Higher correlation strength values ​​result in greater weights for the base curves of adjacent layers. For example, if the correlation strength between the central region feature layer and the transition region feature layer is high, their base curves will have larger weights during fusion. A weighted average is then used to fuse the three base curves into a preliminary luminance mapping curve. This involves weighting the three base curves according to their respective weights. For instance, for the luminance value at each time point, the luminance values ​​of the central region base curve, transition region base curve, and edge region base curve at that time point are multiplied by their respective weights, and the results are summed to obtain the luminance value of the preliminary luminance mapping curve at that time point. Through inter-layer fusion, the feature correlations between different layers can be comprehensively considered, generating a preliminary luminance mapping curve that better meets the needs of human visual perception.

[0102] Step S360: Perform temporal smoothing optimization on the initial brightness mapping curve, calculate the slope change of the curve at adjacent sampling times, and if the change exceeds a preset threshold, adjust the curve value at the next time step so that the slope change meets the smoothing requirements, thereby generating a dynamic brightness mapping curve.

[0103] Temporal smoothing optimization of the initial brightness mapping curve aims to eliminate abrupt changes and fluctuations in the curve, making it smoother. The slope change of the curve at adjacent sampling times is calculated by dividing the difference between the curve values ​​at adjacent sampling times by the time interval to obtain the slope of the curve within that time period. Then, the difference between adjacent slopes is calculated to obtain the slope change.

[0104] If the change exceeds a preset threshold, the curve value at the next time step is adjusted. The preset threshold is a pre-defined value used to determine if the slope change is too large. If the slope change exceeds the preset threshold, it indicates that the curve changes too drastically within that time period, and the curve value at the next time step needs to be adjusted to ensure the slope change meets smoothing requirements. For example, the curve value at the next time step can be adjusted proportionally based on the deviation between the slope change and the threshold.

[0105] In one implementation, step S360 may specifically include the following steps S361 to S366:

[0106] Step S361: Calculate the slope of the preliminary brightness mapping curve by dividing the difference between the curve values ​​at adjacent sampling times by the time interval, and generate a slope sequence.

[0107] The slope calculation of the preliminary brightness mapping curve is performed to analyze the rate of change of the curve at each sampling point. The calculation is done by dividing the difference in curve values ​​between adjacent sampling times by the time interval. For example, for two adjacent sampling times t1 and t2, if the value of the curve at time t1 is y1 and the value at time t2 is y2, and the time interval is Δt, then the slope of the curve at time t1 is (y2-y1) / Δt. Generating a slope sequence involves arranging the calculated slope values ​​for each sampling point in chronological order. Each element of the sequence represents the slope value for the corresponding sampling point. By generating the slope sequence, the slope information of the preliminary brightness mapping curve can be converted into a sequence form that is easy to process and analyze.

[0108] Step S362: Extract the change in the slope sequence, calculate the difference between adjacent slope values, and generate a slope change sequence. A positive change indicates that the slope is increasing, and a negative change indicates that the slope is decreasing.

[0109] Extracting changes in the slope sequence is crucial for analyzing slope variations. The difference between adjacent slope values ​​is calculated to obtain the slope change. For example, for two adjacent slope values ​​k1 and k2, the slope change is k2 - k1. A positive change indicates an increasing slope, while a negative change indicates a decreasing slope. Generating the slope change sequence involves arranging the calculated slope changes in chronological order, where each element represents the change in the corresponding adjacent slope value.

[0110] Step S363: Determine the threshold range of slope change based on the maximum and minimum values ​​of the preliminary brightness mapping curve. The larger the difference between the maximum and minimum values, the wider the threshold range. Generate the upper limit and lower limit of the dynamic threshold.

[0111] Determining the threshold range for slope change based on the maximum and minimum values ​​of the initial brightness mapping curve is to set the threshold for slope change according to the curve's fluctuations. A larger difference between the maximum and minimum values ​​indicates greater curve fluctuation, and a wider threshold range. Generating dynamic upper and lower thresholds involves using the upper and lower limits of the threshold range as the dynamic upper and lower thresholds, respectively. These dynamic upper and lower thresholds are used to determine whether the slope change exceeds a preset threshold. By generating dynamic upper and lower thresholds, the threshold range for slope change can be dynamically adjusted based on the actual situation of the curve.

[0112] Step S364: Traverse the slope change sequence, identify the threshold points where the change exceeds the upper limit of the dynamic threshold or is lower than the lower limit of the dynamic threshold, and record the position index of the threshold point and the corresponding slope change.

[0113] Traversing the slope change sequence involves examining each element in the sequence sequentially. Points where the change exceeds the upper or lower limit of the dynamic threshold are identified. If a slope change exceeds the upper or lower limit, it indicates that the slope change at that point is too drastic, thus exceeding the threshold.

[0114] Recording the location index and corresponding slope change of each threshold-exceeding point is for subsequent processing. The location index indicates the position of the threshold-exceeding point in the slope change sequence, and the slope change indicates the slope change at that point. By identifying threshold-exceeding points and recording their location index and slope change, points requiring adjustment can be identified.

[0115] Step S365: Adjust and calculate the curve value at the next sampling time corresponding to the point exceeding the threshold. Adjust the curve value proportionally according to the degree of deviation between the slope change of the point exceeding the threshold and the threshold. The greater the deviation, the greater the adjustment.

[0116] Adjusting the curve value at the next sampling time corresponding to the threshold point is to ensure the slope change of the curve meets smoothing requirements. The curve value is adjusted proportionally based on the deviation of the slope change at the threshold point from the threshold value. For example, if the slope change at the threshold point significantly exceeds the upper limit of the dynamic threshold, it indicates a large deviation, and the adjustment will be larger; if the slope change slightly exceeds the upper limit of the dynamic threshold, it indicates a small deviation, and the adjustment will be smaller. By adjusting the curve value proportionally, the slope change of the curve can be kept within a reasonable range, thus making the curve smoother.

[0117] Step S366: Recalculate the slope sequence and slope change sequence of the adjusted initial brightness mapping curve, and verify whether all changes are within the dynamic threshold range. If there are points exceeding the threshold, perform the over-threshold point identification and curve value adjustment operation again until all changes meet the smoothness requirements and generate the dynamic brightness mapping curve.

[0118] Recalculating the slope sequence and slope change sequence of the adjusted initial brightness mapping curve is to verify whether the adjusted curve meets the smoothness requirements. It verifies whether all changes are within the dynamic threshold range. If all slope changes are between the upper and lower limits of the dynamic threshold, the curve meets the smoothness requirements; if there are threshold-exceeding points, the threshold-exceeding point identification and curve value adjustment operations are performed again. By continuously identifying threshold-exceeding points and adjusting curve values, the initial brightness mapping curve can eventually become smoother, generating a dynamic brightness mapping curve that meets the requirements.

[0119] Step S400: Call the pre-built partitioned light effect model. The partitioned light effect model includes the light effect output characteristics of each partition LED under different driving conditions and the light crosstalk influence between adjacent partitions. The partitioned light effect model is pre-built based on the temperature sensitivity test data and light crosstalk test data before leaving the factory.

[0120] The purpose of calling the pre-built zone luminous efficacy model is to obtain the luminous efficacy output characteristics of each zone's LEDs under different driving conditions and the degree of optical crosstalk between adjacent zones. The zone luminous efficacy model is pre-built based on temperature sensitivity test data and optical crosstalk test data before leaving the factory. The temperature sensitivity test data reflects the changes in luminous efficacy output of each zone's LEDs at different temperatures; for example, as the temperature increases, the luminous efficacy output of the LEDs may decrease.

[0121] Optical crosstalk test data reflects the degree of optical crosstalk between adjacent zones. Optical crosstalk is the mutual interference between lights in adjacent zones, affecting their respective luminous efficacy output. By calling the pre-built zone luminous efficacy model, the luminous efficacy output characteristics of each zone's LEDs and the degree of optical crosstalk between adjacent zones can be accurately understood, providing a basis for brightness adjustment.

[0122] In one implementation, step S400 may specifically include the following steps S410 to S460:

[0123] Step S410: Call the pre-stored spatial distribution matrix. The spatial distribution matrix is ​​generated based on the spatial coordinate marking of the MiniLED backlight zoned LED cluster before leaving the factory. It records the two-dimensional coordinates of all LEDs in each zoned LED sub-cluster, and the matrix elements are the set of LED coordinates of the corresponding zone.

[0124] The pre-stored spatial distribution matrix is ​​invoked to obtain the spatial coordinate information of the MiniLED backlight zoned LED clusters. This spatial distribution matrix is ​​generated based on the spatial coordinate markings of the MiniLED backlight zoned LED clusters before shipment. Before leaving the factory, the two-dimensional coordinates of all LEDs in each zone's LED sub-cluster are marked, and this coordinate information is then stored in the spatial distribution matrix. Matrix elements are the sets of LED coordinates for the corresponding zone. For example, the first row of the matrix represents the set of LED coordinates for the first zone, and the second row represents the set of LED coordinates for the second zone.

[0125] Step S420: Call the pre-generated temperature sensitivity coefficient sequence. The temperature sensitivity coefficient sequence is generated based on the temperature sensitivity test data of each partition LED sub-cluster before leaving the factory, describing the change law of light effect output under different ambient temperatures.

[0126] The purpose of calling the pre-generated temperature sensitivity coefficient sequence is to obtain the variation pattern of luminous efficacy output of each zone's LED chips under different ambient temperatures. The temperature sensitivity coefficient sequence is generated based on the temperature sensitivity test data of each zone's LED chip sub-cluster before leaving the factory. Before leaving the factory, the luminous efficacy output of each zone's LED chip sub-cluster is tested at different temperatures, and the corresponding temperature and luminous efficacy output data are recorded. Then, the temperature sensitivity coefficient sequence is generated based on this data.

[0127] In one implementation, step S420 may specifically include the following steps S421 to S425:

[0128] Step S421: Read the pre-stored temperature sensitivity coefficient sequence from the display storage unit. The temperature sensitivity coefficient sequence is generated by fitting the light effect of each zone LED sub-cluster under different ambient temperatures before leaving the factory, and contains sensitivity coefficients corresponding to multiple temperature nodes.

[0129] The pre-stored temperature sensitivity coefficient sequence is read from the display storage unit, a device used to store various data, such as memory or a hard drive. The temperature sensitivity coefficient sequence is generated by sampling the luminous efficacy of each LED cluster under different ambient temperatures before shipment. Before shipment, the luminous efficacy output of each LED cluster is sampled under different ambient temperatures, and the corresponding temperature and luminous efficacy output data are recorded. Then, a fitting algorithm is used to process this data to generate the temperature sensitivity coefficient sequence.

[0130] Step S422: Obtain the real-time temperature value collected by the current ambient temperature sensor, match the real-time temperature value with the temperature nodes in the temperature sensitivity coefficient sequence, and if the real-time temperature value matches a certain temperature node, directly call the sensitivity coefficient corresponding to that node.

[0131] The process involves acquiring the real-time temperature value from an ambient temperature sensor, such as a thermocouple or thermistor, used to measure the current ambient temperature. Matching the real-time temperature value with temperature nodes in a temperature sensitivity coefficient sequence checks whether the real-time temperature value matches a specific temperature node in the sequence.

[0132] If the real-time temperature value matches a certain temperature node, the sensitivity coefficient corresponding to that node is directly called. For example, if the real-time temperature value is 25℃, and the temperature sensitivity coefficient sequence contains a temperature node of 25℃, then the sensitivity coefficient corresponding to 25℃ is directly called.

[0133] Step S423: If the real-time temperature value is located between two adjacent temperature nodes, the sensitivity coefficient corresponding to the real-time temperature is calculated by linear interpolation, and the interpolation weight is inversely proportional to the distance between the two nodes.

[0134] If the real-time temperature value lies between two adjacent temperature nodes, it means that the real-time temperature value does not correspond to any temperature node in the temperature sensitivity coefficient sequence. The sensitivity coefficient corresponding to the real-time temperature is calculated using linear interpolation, a method of interpolation between two known points. For example, if the real-time temperature value T lies between temperature nodes T1 and T2, and the corresponding sensitivity coefficients are S1 and S2 respectively, then the sensitivity coefficient S corresponding to the real-time temperature can be calculated using the following formula: S = S1 + (S2 - S1) * (T - T1) / (T2 - T1).

[0135] The interpolation weight is inversely proportional to the distance between the real-time temperature and the two nodes. That is, the closer the real-time temperature is to a node, the greater the weight of the sensitivity coefficient corresponding to that node in the interpolation calculation.

[0136] Step S424: Arrange the sensitivity coefficients obtained by matching or interpolation in partition order to generate a partition temperature sensitivity coefficient vector at the current temperature, which serves as the real-time call result of the temperature sensitivity coefficient sequence.

[0137] Arranging the sensitivity coefficients obtained through matching or interpolation by partition order involves arranging the sensitivity coefficients of each partition into a vector according to the partition number. For example, the sensitivity coefficient of the first partition is placed in the first element of the vector, and the sensitivity coefficient of the second partition is placed in the second element. A partition temperature sensitivity coefficient vector is generated at the current temperature, serving as the real-time result of the temperature sensitivity coefficient sequence. This vector reflects the changes in luminous efficacy output of each partition's LEDs at the current temperature.

[0138] Step S425: Adjust the temperature influence term in the partitioned light effect model based on the partitioned temperature sensitivity coefficient vector so that the light effect output calculated by the partitioned light effect model reflects the actual characteristics at the current temperature.

[0139] Adjusting the temperature influence term in the partitioned luminous efficacy model based on the partitioned temperature sensitivity coefficient vector involves applying the sensitivity coefficients from the partitioned temperature sensitivity coefficient vector to the partitioned luminous efficacy model. The temperature influence term in the partitioned luminous efficacy model describes the effect of temperature on luminous efficacy output. By substituting the sensitivity coefficients from the partitioned temperature sensitivity coefficient vector into the temperature influence term, the calculation results of the partitioned luminous efficacy model can be adjusted to reflect the actual luminous efficacy output characteristics at the current temperature.

[0140] For example, the temperature effect term in the partitioned lighting effect model can be represented as a function whose input is the temperature sensitivity coefficient and whose output is the temperature correction coefficient for the lighting effect output. Substituting the sensitivity coefficients in the partitioned temperature sensitivity coefficient vector into this function yields the correction coefficients for each partition, and these correction coefficients are then applied to the lighting effect output calculation of the partitioned lighting effect model.

[0141] Step S430: Call the pre-built partition crosstalk matrix. The partition crosstalk matrix is ​​generated based on the optical crosstalk test data of adjacent partition LED sub-clusters before leaving the factory, and includes the crosstalk influence between adjacent partitions.

[0142] The purpose of calling the pre-built partition crosstalk matrix is ​​to obtain the optical crosstalk impact degree between adjacent partitions. The partition crosstalk matrix is ​​generated based on optical crosstalk test data of adjacent partition LED sub-clusters before shipment. Before shipment, the optical crosstalk between adjacent partition LED sub-clusters is tested, and the corresponding optical crosstalk data is recorded. The partition crosstalk matrix is ​​then generated based on this data. The partition crosstalk matrix contains the crosstalk impact degree between adjacent partitions. Each element of the matrix represents the degree of crosstalk impact between corresponding adjacent partitions. For example, the element in the first row and second column of the matrix represents the crosstalk impact degree of the first partition on the second partition.

[0143] In one implementation, step S430 may specifically include the following steps S431 to S435:

[0144] Step S431: Read the pre-stored partition crosstalk matrix from the display storage unit. The partition crosstalk matrix is ​​calculated and generated before leaving the factory by controlling the change of driving conditions of a single partition and collecting the change of light effect of adjacent partitions. The rows represent the interfering partitions, the columns represent the interfering partitions, and the elements are the crosstalk influence degree.

[0145] The pre-stored partition crosstalk matrix is ​​read from the display storage unit, a device used to store various data, such as memory or a hard drive. The partition crosstalk matrix is ​​calculated before shipment by controlling changes in the driving conditions of a single partition and collecting data on changes in the luminous efficacy of adjacent partitions. Before shipment, the driving conditions of a certain partition are controlled to change, for example, by altering the current or voltage of the LEDs in that partition. Then, data on changes in the luminous efficacy of adjacent partitions are collected, and the crosstalk influence between adjacent partitions is calculated based on these changes. Rows represent interfering partitions, and columns represent affected partitions. For example, the element in the first row and second column of the matrix represents the crosstalk influence of the first partition on the second partition. The element represents the crosstalk influence, which indicates the degree to which the luminous efficacy output of the interfering partition affects the luminous efficacy output of the affected partition.

[0146] Step S432: Based on the current partition working status of the display screen, identify the set of partitions in the active state, set the crosstalk influence of inactive partitions to zero, and generate the crosstalk submatrix of active partitions.

[0147] Based on the current operating status of the display's partitions, identify the set of partitions that are active. The current operating status of the display's partitions can be determined by detecting the drive signals of the LEDs in each partition. If a drive signal is input to the LEDs in a certain partition, it means that the partition is active; if no drive signal is input, it means that the partition is inactive.

[0148] The crosstalk impact of inactive partitions is set to zero because inactive partitions have no luminous effect output and will not cause crosstalk to other partitions. Generating an active partition crosstalk submatrix involves extracting the rows and columns corresponding to the active partitions from the partition crosstalk matrix and forming a new matrix. For example, if partitions 1, 3, and 5 are active, then the first, third, and fifth rows and columns of the partition crosstalk matrix are extracted to form the active partition crosstalk submatrix. Generating this active partition crosstalk submatrix reduces unnecessary calculations and improves the efficiency of luminous effect calculation and control.

[0149] Step S433: Sparsify the active partition crosstalk submatrix, retaining elements whose absolute value is greater than the preset crosstalk threshold, and setting the remaining elements to zero to generate a sparse crosstalk matrix.

[0150] The preset crosstalk threshold is a pre-defined value used to determine whether the crosstalk influence represented by an element is sufficiently significant. When the absolute value of an element in the active partition crosstalk submatrix is ​​greater than the preset crosstalk threshold, it indicates that the crosstalk influence between adjacent partitions corresponding to that element is relatively significant and needs to be retained; conversely, when the absolute value of an element is less than or equal to the preset crosstalk threshold, the crosstalk influence is considered negligible and is set to zero. For example, in a certain display screen, setting the preset crosstalk threshold to a relatively small value and scanning the active partition crosstalk submatrix reveals elements with low crosstalk influence, such as those between adjacent partitions where crosstalk is weak due to distance or structural design, which are set to zero. This sparsity processing further reduces the complexity of subsequent calculations, improves computational efficiency, and does not substantially affect the overall light effect calculation results.

[0151] Step S434: Map the row and column indices of the sparse crosstalk matrix to the current partition number so that the partition positions corresponding to the matrix elements are consistent with the actual display screen partition layout, and generate the partition crosstalk matrix after position calibration.

[0152] After the preceding processing, the row and column indices of the sparse crosstalk matrix may not directly correspond to the actual partition numbers of the display screen. Therefore, a mapping operation is required to ensure that the matrix elements accurately reflect the actual partition position relationships. Specifically, a mapping table can be established to record the correspondence between the row and column indices of the sparse crosstalk matrix and the actual partition numbers. For example, based on the physical layout and partition numbering rules of the display screen, the first row of the sparse crosstalk matrix can be mapped to the third partition of the actual display screen, the second row to the fifth partition, and so on. Through this mapping operation, the sparse crosstalk matrix is ​​converted into a position-calibrated partition crosstalk matrix, ensuring that the crosstalk influence represented by the matrix elements matches the position and relationship of adjacent partitions in the actual display screen. This provides a foundation for accurately calculating the impact of changes in the driving conditions of adjacent partitions on the light output of the current partition.

[0153] Step S435: Calculate the impact of changes in the driving conditions of adjacent partitions on the light output of the current partition based on the partition crosstalk matrix after position calibration.

[0154] When calculating the impact of changes in the driving conditions of adjacent zones on the luminous efficacy output of the current zone, it is necessary to consider both the driving conditions and luminous efficacy output characteristics of each zone. Specifically, for each zone, the impact of adjacent zones on its luminous efficacy output is calculated based on changes in the driving conditions of its adjacent zones and the corresponding crosstalk influence in the crosstalk matrix after position calibration. For example, when the driving current of an adjacent zone increases, the increase or decrease in the luminous efficacy output of the current zone is calculated based on the corresponding crosstalk influence in the crosstalk matrix. Through this calculation, the impact of changes in the driving conditions of adjacent zones on the luminous efficacy output of the current zone can be accurately assessed, providing a basis for luminous efficacy control to ensure that the luminous efficacy output of each zone of the display screen meets expectations and to avoid poor display effects caused by crosstalk.

[0155] Step S440: Sample the dynamic brightness mapping curve for each zone's brightness requirement and generate a target brightness requirement sequence for each zone.

[0156] The dynamic brightness mapping curve describes the non-linear transformation relationship from input brightness value to output brightness value, comprehensively considering the dynamic changes in ambient light, the content features of the currently displayed image, and the visual perception requirements of the human eye. Zoned brightness requirement sampling divides the dynamic brightness mapping curve according to the zones of the display screen. For each zone, the brightness value at the corresponding position on the dynamic brightness mapping curve is extracted. In practice, the sampling point on the dynamic brightness mapping curve can be determined based on the position and range of each zone on the display screen. For example, for the zone in the upper left corner of the display screen, the corresponding position on the dynamic brightness mapping curve is found, and the brightness value at that position is read as the target brightness requirement for that zone. By performing zoned brightness requirement sampling on the dynamic brightness mapping curve, the brightness mapping relationship of the entire display screen is refined to each zone, generating a target brightness requirement sequence for each zone. Each element in this sequence corresponds to a target brightness value for a zone, providing a clear target for subsequent light effect model construction and driving parameter calculation.

[0157] Step S450: Connect the spatial distribution matrix, temperature sensitivity coefficient sequence, partition crosstalk matrix and target brightness requirement sequence to generate a partitioned luminous effect model that includes luminous effect output characteristics and crosstalk effects.

[0158] Correlation modeling involves comprehensively considering the interrelationships between various factors to construct a model that can accurately describe the light output of each zone of the display screen.

[0159] In one implementation, step S450 may specifically include the following steps S451 to S456:

[0160] Step S451: Decompose the target brightness demand sequence into a basic brightness demand component and a crosstalk compensation demand component.

[0161] The target luminance requirement sequence represents the desired luminance values ​​for each zone. However, actual luminous efficacy output is affected by various factors, with crosstalk being a significant one. Therefore, the target luminance requirement needs to be decomposed to separately handle the base luminance and crosstalk compensation. The base luminance requirement component is the luminance value required by each zone to achieve the target luminance without considering crosstalk. The crosstalk compensation requirement component is the additional luminance value required to offset the crosstalk effect of adjacent zones, enabling each zone to ultimately achieve its target luminance. For example, in a given zone, crosstalk from adjacent zones may increase or decrease its luminous efficacy output. To ensure that zone reaches its target luminance, the corresponding crosstalk compensation requirement component needs to be determined based on the degree and direction of the crosstalk effect.

[0162] Step S452: Construct a basic luminous efficacy model based on the spatial distribution matrix and temperature sensitivity coefficient sequence. The model input is the driving parameters, and the output is the basic luminous efficacy value without crosstalk. Input the basic brightness requirement component into the basic luminous efficacy model to generate basic driving condition parameters.

[0163] The spatial distribution matrix records the spatial coordinates of the LEDs in each zone, while the temperature sensitivity coefficient sequence describes the variation in luminous efficacy output of the LEDs under different ambient temperatures. Combining these two factors, a basic luminous efficacy model can be constructed. This model reflects the relationship between the luminous efficacy output of the LEDs in each zone and the driving parameters under crosstalk-free conditions. Driving parameters can include current, voltage, etc., and adjusting these parameters controls the luminous efficacy output of the LEDs. By inputting the basic brightness requirement component into the basic luminous efficacy model, the model calculates the required driving parameter values ​​for each zone to achieve the basic brightness requirement, i.e., the basic driving condition parameters, based on its internal mapping relationship. For example, for a given zone, based on its basic brightness requirement component, the basic luminous efficacy model calculates the corresponding driving current or voltage value to ensure that the zone can achieve the desired basic luminous efficacy output under crosstalk-free conditions.

[0164] Step S453: Input the basic driving condition parameters into the basic luminous efficacy model to obtain the basic luminous efficacy output value of each partition under the condition of no crosstalk. Input the basic luminous efficacy output value into the partition crosstalk matrix and calculate the crosstalk luminous efficacy prediction value of the adjacent partition to the current partition. The prediction value is the sum of the product of the basic luminous efficacy output value of the adjacent partition and the corresponding crosstalk influence degree.

[0165] After inputting the basic driving condition parameters into the basic luminous efficacy model, the model outputs the basic luminous efficacy output values ​​of each partition under crosstalk-free conditions based on its internal calculation logic. These basic luminous efficacy output values ​​reflect the luminous efficacy performance of each partition under ideal conditions. Then, these basic luminous efficacy output values ​​are input into the partition crosstalk matrix. Based on the crosstalk influence degree between adjacent partitions recorded in the matrix, the predicted crosstalk luminous efficacy value of the current partition by the adjacent partitions is calculated. Specifically, for each partition, the basic luminous efficacy output values ​​of its adjacent partitions are multiplied by the corresponding crosstalk influence degree, and then these products are added together to obtain the predicted crosstalk luminous efficacy value of that partition. For example, if a partition has two adjacent partitions, the basic luminous efficacy output value of adjacent partition A is a certain value, and its crosstalk influence degree on that partition is a specific value. The basic luminous efficacy output value and crosstalk influence degree of adjacent partition B are also known. By multiplying the basic luminous efficacy output value of adjacent partition A by its crosstalk influence degree, and the basic luminous efficacy output value of adjacent partition B by its crosstalk influence degree, and then adding these two products together, the predicted crosstalk luminous efficacy value of that partition is obtained.

[0166] Step S454: Calculate the crosstalk compensation driving parameters based on the predicted crosstalk light effect and the crosstalk compensation requirement component. The magnitude of the compensation parameters is proportional to the predicted crosstalk value, and the direction is opposite to the direction of crosstalk influence.

[0167] The predicted crosstalk luminous efficacy value reflects the degree and direction of the impact of crosstalk from adjacent zones on the luminous efficacy output of the current zone. The crosstalk compensation requirement component is the additional luminance value required to offset this crosstalk effect. Based on these two values, the crosstalk compensation drive parameter can be calculated. Since the magnitude of the compensation parameter is proportional to the predicted crosstalk value (i.e., the larger the predicted crosstalk value, the larger the required compensation drive parameter), and its direction is opposite to the direction of the crosstalk effect, the compensation drive parameter is negative when crosstalk increases the luminous efficacy output of the current zone to reduce the luminous efficacy output; conversely, it is positive when crosstalk decreases the luminous efficacy output of the current zone to increase the luminous efficacy output. For example, if the predicted crosstalk luminous efficacy value shows that crosstalk from adjacent zones increases the luminous efficacy output of the current zone by a certain amount, then the crosstalk compensation drive parameter will be negative. By adjusting the drive parameter, the luminous efficacy output of that zone is reduced to offset the crosstalk effect. By calculating the crosstalk compensation drive parameter, the crosstalk effect of adjacent zones can be effectively compensated, making the luminous efficacy output of each zone closer to the target luminance.

[0168] Step S455: Vector superposition of the basic driving condition parameters and crosstalk compensation driving parameters to generate comprehensive driving condition parameters. The superposition weight is determined according to the partition visual weight label.

[0169] The basic driving condition parameters are the driving parameters required to ensure that each zone achieves the basic brightness requirement without crosstalk. The crosstalk compensation driving parameters are the additional driving parameters required to offset the crosstalk effect between adjacent zones. These two parameters are vector-superimposed to obtain the comprehensive driving condition parameters. During the superposition process, the superposition weight is determined based on the visual weight label of each zone. The visual weight label reflects the human eye's attention to and importance of different zones. For zones with high human eye attention, such as the center area of ​​the display screen, the superposition weight is relatively large, meaning that the basic driving condition parameters and crosstalk compensation driving parameters have a more significant impact on the calculation of the comprehensive driving condition parameters. Conversely, for zones with low human eye attention, such as the edge areas of the display screen, the superposition weight is relatively small. For example, in a certain zone, the superposition weight determined by its visual weight label is a specific value. The basic driving condition parameters are multiplied by this weight, and the crosstalk compensation driving parameters are also multiplied by this weight. Then, the two are added together to obtain the comprehensive driving condition parameters for that zone. By determining the superposition weight based on the visual weight label of each zone, the basic brightness requirement and the crosstalk compensation requirement can be considered more reasonably, generating comprehensive driving condition parameters that better meet the visual perception requirements of the human eye.

[0170] Step S456: Input the comprehensive driving condition parameters into the basic luminous efficacy model, calculate the actual luminous efficacy output value, compare it with the target brightness requirement sequence to obtain the model deviation. If the deviation is less than the preset threshold, then encapsulate the basic luminous efficacy model, the partition crosstalk matrix and the comprehensive driving parameter calculation logic into a partition luminous efficacy model.

[0171] After inputting the comprehensive driving condition parameters into the basic luminous efficacy model, the model outputs the actual luminous efficacy output values ​​for each zone based on its internal calculation logic. These actual luminous efficacy output values ​​are compared with the target brightness values ​​in the target brightness requirement sequence, and the difference between the two is calculated, i.e., the model deviation. A preset threshold is a pre-defined allowable deviation range. If the model deviation is less than the preset threshold, it indicates that the calculated actual luminous efficacy output value is close to the target brightness requirement, and the calculation result of the zone luminous efficacy model can be considered accurate and effective. The basic luminous efficacy model, the zone crosstalk matrix, and the comprehensive driving parameter calculation logic are encapsulated to form a complete zone luminous efficacy model. This model can comprehensively consider the luminous efficacy output characteristics of each zone, the crosstalk effect of adjacent zones, and the visual perception requirements of the human eye, providing an accurate calculation basis for subsequent brightness control. If the model deviation is greater than the preset threshold, the comprehensive driving condition parameters need to be adjusted and recalculated until the model deviation is less than the preset threshold.

[0172] Based on the above, the partitioned luminous efficacy model is composed of multiple sub-models and data structures, including a basic luminous efficacy sub-model. This sub-model is built upon a spatial distribution matrix and a temperature sensitivity coefficient sequence. The spatial distribution matrix records the spatial coordinates of the LEDs in each partition, reflecting their physical layout on the display screen. Different layouts affect light propagation and distribution. The temperature sensitivity coefficient sequence describes the variation of LED luminous efficacy output under different ambient temperatures, as the luminous efficacy changes with temperature; for example, increased temperature may lead to decreased luminous efficacy. The input to the basic luminous efficacy sub-model is driving parameters (such as current and voltage), and the output is the basic luminous efficacy value without crosstalk. This value can be represented by a nonlinear function, such as a polynomial function or a neural network model, to calculate the luminous efficacy output of each partition under ideal conditions based on the driving parameters, spatial distribution, and temperature effects. The partitioned luminous efficacy model also includes a crosstalk sub-model, built upon a partitioned crosstalk matrix, which records the degree of crosstalk influence between adjacent partitions. The sub-model calculates the predicted crosstalk luminous efficacy of adjacent zones on the current zone. This is achieved by multiplying and summing the basic luminous efficacy output values ​​of each zone with the corresponding crosstalk influence in the zone crosstalk matrix. The zone luminous efficacy model includes a comprehensive calculation module that integrates the results of the basic luminous efficacy sub-model and the crosstalk sub-model. First, it decomposes the target brightness demand sequence into a basic brightness demand component and a crosstalk compensation demand component. Then, it calculates the basic driving condition parameters based on the basic brightness demand component and combines them with the predicted crosstalk luminous efficacy values ​​to calculate the crosstalk compensation driving parameters. Finally, it vector-superimposes the basic driving condition parameters and the crosstalk compensation driving parameters to generate comprehensive driving condition parameters, with the superposition weights determined based on the zone visual weight labels. Through this comprehensive calculation process, the zone luminous efficacy model can output actual luminous efficacy output values ​​that consider crosstalk influence and temperature factors.

[0173] Step S460: Adapt the parameters of the partitioned light effect model. Based on the current ambient temperature, query the corresponding coefficient in the temperature sensitivity coefficient sequence and adjust the temperature influence parameters in the model so that the model output adapts to the current temperature conditions.

[0174] While the zonal luminous efficacy model considers the impact of temperature on luminous efficacy output during its construction, ambient temperature is dynamic. Therefore, parameter adaptation of the zonal luminous efficacy model is necessary to ensure that the model output accurately reflects the luminous efficacy under the current temperature conditions. Based on the current ambient temperature, the corresponding coefficient in the temperature sensitivity coefficient sequence is consulted. This coefficient reflects the proportion of change in LED luminous efficacy output at the current temperature relative to the luminous efficacy output at the standard temperature. This coefficient is then applied to the temperature influence parameter in the zonal luminous efficacy model to adjust the model. For example, if the current ambient temperature is higher than the standard temperature, the corresponding coefficient found in the temperature sensitivity coefficient sequence indicates that the LED luminous efficacy output will decrease. Therefore, the temperature influence parameter in the zonal luminous efficacy model is adjusted accordingly to account for this decrease in the calculated luminous efficacy output value. Through this parameter adaptation operation, the zonal luminous efficacy model can dynamically adjust its output according to the current ambient temperature, improving the model's accuracy and adaptability, and ensuring that the display screen maintains good display performance under different temperature environments.

[0175] Step S500: Based on the partition light effect model and dynamic brightness mapping curve, perform reverse solution to generate partition calibration driving parameters, and send the partition calibration driving parameters to the drive control unit of the MiniLED backlight module to drive brightness adjustment.

[0176] The zone luminous efficacy model describes the relationship between the luminous efficacy output of each zone and the driving parameters, while the dynamic luminance mapping curve specifies the target luminance value for each zone. The purpose of inverse solving is to calculate, based on these two conditions, the driving parameter values ​​required to make each zone achieve the target luminance, i.e., the zone calibration driving parameters.

[0177] In one implementation, step S500 may specifically include the following steps S510-S560:

[0178] Step S510: Extract the target brightness value of each partition from the dynamic brightness mapping curve, find the corresponding sampling point on the curve according to the partition number, read the brightness value of the sampling point as the target brightness of the partition, and generate a target brightness vector. The order of the vector elements corresponds to the partition number.

[0179] The dynamic brightness mapping curve, obtained after a series of processing steps, comprehensively considers factors such as ambient light, image content, and human visual perception, providing appropriate target brightness for each zone. Based on the zone number, the corresponding sampling point is found on the dynamic brightness mapping curve; the brightness value of each sampling point is the target brightness for that zone. The target brightness values ​​of each zone are arranged in order of zone number to generate a target brightness vector. For example, if the display screen has three zones, numbered 1, 2, and 3, the target brightness value extracted from the dynamic brightness mapping curve for zone 1 is one specific value, for zone 2 it's another, and for zone 3 it's yet another. These three values ​​are then arranged sequentially into a vector, which is the target brightness vector.

[0180] Step S520: Call the forward calculation interface of the partitioned light effect model, input the driving parameter vector, and output the light effect output vector. This interface realizes the mapping between the light effect output and the driving parameters based on the pre-configured calculation logic, so that the calculation process can adapt to the current ambient temperature and partitioned crosstalk state.

[0181] The forward calculation interface of the partitioned lighting effect model is a pre-configured calculation module that can calculate the lighting effect output value of each partition based on the input driving parameter vector, forming a lighting effect output vector. The calculation logic of this interface considers the current ambient temperature and partition crosstalk state, ensuring that the calculation results accurately reflect the actual situation by calling information such as the temperature sensitivity coefficient sequence and the partition crosstalk matrix. For example, given an input driving parameter vector containing the driving parameter values ​​of each partition, the forward calculation interface will calculate the lighting effect output value of each partition based on the internal calculation logic of the partitioned lighting effect model, combined with the temperature influence parameters under the current ambient temperature and the influence of partition crosstalk, and then assemble these values ​​into a lighting effect output vector.

[0182] Step S530: Call the pre-built reverse solution objective function. The reverse solution objective function uses the driving parameter vector as the variable. The objective function value is the sum of squares of the deviations between the light effect output vector and the target brightness vector. The function parameters are optimized and calibrated before leaving the factory.

[0183] The pre-built inverse solving objective function is designed to find the driving parameters that make the luminous effect output of each zone as close as possible to the target brightness. This function uses the driving parameter vector as a variable and continuously adjusts the values ​​of the driving parameter vector to minimize the sum of squared deviations between the luminous effect output vector and the target brightness vector. Calculating the sum of squared deviations amplifies the impact of deviations, making the objective function focus more on larger deviations, thus more effectively finding the optimal driving parameters. The function parameters are optimized and calibrated before leaving the factory. Before leaving the factory, through extensive experiments and optimizations, the optimal parameter values ​​of the objective function are determined and fixed to ensure the accuracy and stability of the inverse solution. For example, during the inverse solution process, the values ​​of the driving parameter vector are continuously changed, and the sum of squared deviations between the corresponding luminous effect output vector and the target brightness vector is calculated. The goal is to find the driving parameter vector that minimizes this sum of squared deviations. By calling the pre-built inverse solving objective function, the inverse solving problem can be transformed into an optimization problem, which can be solved using a suitable algorithm.

[0184] Step S540: Call the pre-stored driver parameter range matrix. The driver parameter range matrix records the upper and lower limits of the driver parameters for each partition. It is pre-configured based on the rated parameters of the LED beads and the safety threshold. The rows represent partitions and the columns represent the upper and lower limits.

[0185] The driving parameter range matrix is ​​used to ensure that the driving parameter values ​​obtained during reverse engineering are within a safe and reasonable range. This matrix records the upper and lower limits of the driving parameters for each partition, which are pre-configured based on the rated parameters and safety thresholds of the LEDs. For example, the rated current and voltage of an LED have certain ranges; exceeding these ranges may damage the LED or affect its lifespan. Therefore, during reverse engineering, the driving parameters need to be limited to these ranges. The rows of the matrix represent partitions, and the columns represent the upper and lower limits. For example, the element in the first row and first column represents the upper limit of the driving parameters for the first partition, and the element in the first row and second column represents the lower limit. By calling the pre-stored driving parameter range matrix, the driving parameters can be constrained during reverse engineering, ensuring the safety and reliability of the solution results.

[0186] Step S550: Call the pre-trained fast solver to solve the inverse objective function, and output the driving parameter vector that minimizes the objective function value under the constraint of the driving parameter range matrix.

[0187] The pre-trained fast solver is a specially trained algorithm module that can quickly and efficiently solve the inverse objective function and find the driving parameter vector that minimizes the objective function value, while satisfying the driving parameter range matrix constraint.

[0188] In one implementation, step S550 may specifically include the following steps S551 to S556:

[0189] Step S551: Input the target brightness vector and the driving parameter range matrix into the initialization module of the fast solver, load the pre-stored initial driving parameter vector, which is pre-configured based on the historical best solution and serves as the starting point for the solution.

[0190] The initialization module of the fast solver receives the target brightness vector and the driving parameter range matrix, and loads a pre-stored initial driving parameter vector. The target brightness vector provides target information for the solution, and the driving parameter range matrix sets the constraints. The initial driving parameter vector is pre-configured based on historical best solutions, representing relatively optimal driving parameter values ​​obtained in previous solutions. Using it as the starting point reduces the search space and improves efficiency. For example, in the initialization module, after inputting the target brightness vector and the driving parameter range matrix, the pre-stored initial driving parameter vector is loaded. Each element in this vector corresponds to the initial driving parameter value for a partition, and the solution process will begin searching from this initial state.

[0191] Step S552: Call the gradient descent unit of the fast solver to calculate the gradient vector of the objective function at the initial driving parameter vector, with the gradient direction pointing in the direction of decreasing function value.

[0192] The gradient descent unit is used to compute the gradient vector of the inversely solved objective function at the current driving parameter vector. The gradient vector represents the rate of change and direction of change of the objective function at that point, with the gradient direction pointing in the direction that the function value decreases. By computing the gradient vector, we can determine in which direction the driving parameters should be adjusted based on the current driving parameter vector to make the objective function value decrease more quickly. For example, at the initial driving parameter vector, the gradient descent unit computes the gradient vector of the objective function at that point according to the expression of the inversely solved objective function and the differentiation rules. Each element of this vector represents the rate of change and direction of change of the corresponding driving parameter.

[0193] Step S553: ​​Update the driving parameter vector according to the gradient vector and the preset step size parameter. The new vector is the initial vector minus the product of the gradient vector and the step size parameter, so that the updated vector is within the range matrix of the driving parameter.

[0194] The driving parameter vector can be updated based on the calculated gradient vector and the preset step size parameter. The preset step size parameter controls the magnitude of each update; a larger step size parameter results in a larger update magnitude, but may cause the solution process to skip the optimal solution; a smaller step size parameter results in a smaller update magnitude, leading to a more stable solution process, but may slow down the solution speed. The updated driving parameter vector is the initial vector minus the product of the gradient vector and the step size parameter, i.e., adjusted in the opposite direction of the gradient to reduce the objective function value. Simultaneously, it is necessary to ensure that the updated vector is within the driving parameter range matrix. If the updated driving parameter value for a certain partition exceeds its upper or lower limit, it is adjusted to the upper or lower limit value. For example, for the driving parameters of a certain partition, the updated value is calculated based on the gradient vector and step size parameter. If this updated value exceeds the upper limit value of that partition in the driving parameter range matrix, it is adjusted to the upper limit value. Through this update method, the driving parameter vector that minimizes the objective function value can be gradually approximated.

[0195] Step S554: Calculate the objective function value corresponding to the updated driving parameter vector and compare it with the function value before the update. If the function value decreases, accept the update; otherwise, reduce the step size parameter and update again.

[0196] Calculate the objective function value corresponding to the updated driving parameter vector and compare it with the objective function value before the update. If the updated objective function value decreases, it means the updated driving parameter vector is closer to the optimal solution, and the update is accepted. If the objective function value does not decrease, it means the update magnitude may be too large, and the step size parameter needs to be reduced. Then, the driving parameter vector is updated again based on the new step size parameter and gradient vector. For example, if the calculated objective function value is smaller than before the update, then the update is accepted, and the updated driving parameter vector is used as the new current vector. If the objective function value does not decrease, the step size parameter is reduced by a certain proportion, such as half of the original value, and then the update operation is performed again. By comparing and adjusting the step size parameter in this way, the stability and effectiveness of the solution process can be guaranteed, avoiding getting trapped in local optima or oscillations.

[0197] Step S555: Repeat the gradient calculation, parameter update and function value comparison steps until the function value change is less than the convergence threshold for a preset number of consecutive iterations, or the maximum number of iterations is reached.

[0198] The process involves repeatedly performing gradient calculations, parameter updates, and function value comparisons, continuously adjusting the driving parameter vector to gradually decrease the objective function value. A convergence threshold, a pre-set small value, is used to determine whether the objective function value has converged to a stable value. When the change in function value for a preset number of consecutive iterations is less than the convergence threshold, it indicates that the objective function value has stabilized, and the solution process can be considered to have converged to the optimal solution. The maximum number of iterations is an upper limit set to prevent the solution process from falling into an infinite loop. If the maximum number of iterations is reached, the solution process stops even if the objective function value has not fully converged. For example, a preset threshold is considered convergent when the change in function value is less than the convergence threshold for 5 consecutive iterations. If the change in the objective function value calculated in the solution process is less than the convergence threshold for 5 consecutive iterations, the iteration stops; if the maximum number of iterations is reached, the iteration also stops. Through this iterative process, the driving parameter vector that minimizes the objective function value can be gradually found.

[0199] Step S556: Output the current driving parameter vector as the solution result. If convergence is not achieved after reaching the maximum number of iterations, call the pre-stored backup driving parameter vector to make the output driving parameters meet the safety requirements.

[0200] When the solution process meets the convergence condition or reaches the maximum number of iterations, the current driving parameter vector is output as the solution result. If convergence is not achieved even after reaching the maximum number of iterations, it indicates that the solution process may have encountered difficulties and cannot find the optimal solution. In this case, a pre-stored backup driving parameter vector is invoked. This backup driving parameter vector is pre-configured based on safety and reliability considerations, ensuring the normal operation of the display screen and avoiding problems caused by unreasonable driving parameters. For example, if the objective function value still has not converged to within the convergence threshold after reaching the maximum number of iterations, the pre-stored backup driving parameter vector is output to ensure that the output driving parameters meet safety requirements, enabling the display screen to operate stably.

[0201] Step S560: Perform partition verification on the driver parameter vector to ensure that each partition parameter generates partition calibration driver parameters that meet hardware safety requirements within the driver parameter range matrix. Perform partition verification on the obtained driver parameter vector to check whether the driver parameter value of each partition is within the upper and lower limits specified in the driver parameter range matrix.

[0202] If the driving parameter value of a certain partition exceeds the range, it is adjusted to the upper or lower limit value to ensure that the parameters of each partition meet the hardware safety requirements. For example, if the driving current value of a certain partition in the driving parameter vector exceeds its upper limit value, the current value is adjusted to the upper limit value. Through partition verification, it can be ensured that the generated partition calibration driving parameters will not damage the hardware, while also ensuring the display effect and stability of the screen. Finally, the partition-verified driving parameter vector is sent to the driving control unit of the MiniLED backlight module as the partition calibration driving parameters. The driving control unit adjusts the brightness of the LEDs in each partition according to these parameters, thereby realizing the brightness control of the MiniLED display based on multi-source environmental perception, so that the display can provide a display effect that meets the visual perception needs of the human eye under different ambient light and image content conditions.

[0203] It is understood that the various algorithms involved in the above descriptions of the embodiments of the present invention, such as Euclidean distance algorithm, cosine distance algorithm, conflict resolution algorithm, etc., can all be obtained from relevant content in the prior art. In order to save space, they will not be elaborated on in the embodiments of the present invention. Furthermore, those skilled in the art can supplement the details when implementing the solution of the present invention based on common knowledge in the art. For example, based on common knowledge in the art, normalization can be used to eliminate dimensional conflicts before feature fusion. For example, in step S210, the fluctuation amplitude feature can be standardized and converted into a dimensionless fluctuation intensity index, and the duration feature can be normalized and converted into a dimensionless duration index. Then, a temporal environmental feature sequence can be generated based on the fluctuation intensity index and the duration index. For another example, in step S260, the brightness sensitivity component, visual comfort component, and trend feature vector can be standardized to convert each component and vector element into a dimensionless index. Then, the standardized brightness sensitivity component, visual comfort component, and trend feature vector can be dynamically assembled and sorted according to regional visual weights to construct a perceptual demand feature set. In step S330, the central region feature layer, The feature values ​​in the temporal feature sequences of the transition region feature layer and the edge region feature layer are standardized and converted into dimensionless index values. Based on the standardized index values, the correlation strength between the central region feature layer and the transition region feature layer, and the correlation strength between the transition region feature layer and the edge region feature layer are calculated. The correlation strength is represented by the ratio of the sum of the products of the corresponding layer index values ​​to the sum of the squares of the corresponding layer index values, generating an inter-layer correlation strength value. In step S340, based on the standardized brightness sensitivity index and visual comfort preference index in the temporal feature sequences of each layer, a brightness mapping base curve for the corresponding region can be generated. The base curve for the central region is dominated by the standardized brightness sensitivity index, the base curve for the edge region is dominated by the standardized visual comfort preference index, and the base curve for the transition region combines both. In step S540, the objective function value is the sum of squares of the deviations between the normalized luminous efficacy output vector and the normalized target brightness vector. Similar content is not elaborated here; those skilled in the art can overcome related problems based on the actual situation and common knowledge in the field.

[0204] In addition, those skilled in the art can also use interpolation to eliminate dimensional differences, combine historical data, experience or business scenario requirements to reasonably set thresholds, train the model based on general model training methods, set the number of layers in the model structure based on actual needs, select activation functions, etc. This invention will not provide redundant descriptions of overly detailed implementation processes.

[0205] Based on the foregoing embodiments, this invention provides a minimumed display screen brightness control device. The various units and modules included in the device can be implemented by a processor in a computer device; of course, they can also be implemented by specific logic circuits. In the implementation process, the processor can be a central processing unit (CPU), a microprocessor unit (MPU), a digital signal processor (DSP), or a field programmable gate array (FPGA), etc.

[0206] Figure 2 This is a schematic diagram of the composition structure of a minimized display screen brightness control device provided in an embodiment of the present invention, as shown below. Figure 2 As shown, the minimized display screen brightness control device 200 includes:

[0207] Information acquisition module 210 is used to acquire ambient light dynamic change information under the current environment and content feature information of the image currently displayed on the display screen. The ambient light dynamic change information includes the light intensity distribution at different times, and the content feature information includes the brightness distribution and color distribution of each display area in the image.

[0208] The demand analysis module 220 is used to input the ambient light dynamic change information and the content feature information into a preset human eye visual perception model to analyze the perception demand and obtain a set of perception demand features that matches the current viewing scene. The set of perception demand features includes the human eye's brightness sensitivity index and visual comfort preference index for different display areas.

[0209] The relationship construction module 230 is used to call the dynamic mapping mechanism to construct the brightness mapping relationship according to the perception requirement feature set, and generate a dynamic brightness mapping curve that reflects the synergistic effect of ambient light and image content. The dynamic brightness mapping curve is used to describe the nonlinear conversion relationship from input brightness value to output brightness value.

[0210] The model calling module 240 is used to call a pre-built partitioned light effect model. The partitioned light effect model includes the light effect output characteristics of each partitioned LED under different driving conditions and the light crosstalk influence between adjacent partitions. The partitioned light effect model is pre-built based on temperature sensitivity test data and light crosstalk test data before leaving the factory.

[0211] The drive adjustment module 250 is used to perform inverse solving based on the partition light effect model and the dynamic brightness mapping curve to generate partition calibration drive parameters, and send the partition calibration drive parameters to the drive control unit of the MiniLED backlight module to drive brightness adjustment.

[0212] The descriptions of the apparatus embodiments above are similar to those of the method embodiments above, and have similar beneficial effects. In some embodiments, the functions or modules included in the apparatus provided by the present invention can be used to perform the methods described in the method embodiments above. For technical details not disclosed in the apparatus embodiments of the present invention, please refer to the descriptions of the method embodiments of the present invention for understanding.

[0213] Figure 3 This is a schematic diagram of a hardware entity of a display screen provided in an embodiment of the present invention, such as... Figure 3 As shown, the hardware entity of the display screen 1000 includes a processor 1001 and a memory 1002, wherein the memory 1002 stores a computer program that can run on the processor 1001, and the processor 1001 executes the program to implement the steps in the method of any of the above embodiments.

Claims

1. A method for brightness control of a minimized display screen based on multi-source environmental perception, characterized in that, The method includes: The system acquires dynamic change information of ambient light under the current environment and content feature information of the image currently displayed on the screen. The dynamic change information of ambient light includes the distribution of light intensity at different times, and the content feature information includes the brightness distribution and color distribution of each display area in the image. The ambient light dynamic change information and the content feature information are input into a preset human eye visual perception model to analyze the perception requirements and obtain a set of perception requirements features that match the current viewing scene. The set of perception requirements features includes the human eye's brightness sensitivity index and visual comfort preference index for different display areas. Based on the set of perceived requirements features, the dynamic mapping mechanism is invoked to construct the brightness mapping relationship, generating a dynamic brightness mapping curve that reflects the synergistic effect between ambient light and image content. The dynamic brightness mapping curve is used to describe the nonlinear conversion relationship from input brightness value to output brightness value. The pre-built partitioned light effect model is invoked. The partitioned light effect model includes the light effect output characteristics of each partition LED under different driving conditions and the light crosstalk influence between adjacent partitions. The partitioned light effect model is pre-built based on temperature sensitivity test data and light crosstalk test data before leaving the factory. Based on the partitioned light effect model and the dynamic brightness mapping curve, the partitioned calibration driving parameters are generated by inverse solution and sent to the drive control unit of the MiniLED backlight module to drive brightness adjustment.

2. The method according to claim 1, characterized in that, The step involves inputting the dynamic change information of ambient light and the content feature information into a preset human visual perception model to analyze perception requirements, thereby obtaining a set of perception requirement features that matches the current viewing scene, including: The ambient light dynamic change information is sampled in the time domain, and divided into a first fluctuation segment and a second fluctuation segment according to the light intensity change cycle. The fluctuation amplitude feature and duration feature of each segment are extracted to generate a time domain environmental feature sequence. The image display area in the content feature information is labeled with regional visual weights, the first type of region and the second type of region in the image are identified, the brightness distribution density and color transition smoothness of each region are calculated respectively, and a regional content feature matrix is ​​generated. The temporal environmental feature sequence is cross-dimensionally correlated with the regional content feature matrix to calculate the response correlation degree between the first fluctuation segment and the first type of region, and the adaptation correlation degree between the second fluctuation segment and the second type of region, thereby generating a cross-sensing correlation matrix. Based on the cross-perception association matrix, a preset visual perception threshold library is queried to obtain the brightness perception threshold and color perception threshold of each region under the current ambient light fluctuation. The dynamic range of each threshold is adjusted according to the region content feature matrix to generate a dynamic perception threshold sequence. The dynamic perception threshold sequence is analyzed hierarchically. First, the brightness sensitivity component of the first type of region is analyzed, and then the visual comfort component of the second type of region is analyzed. The changing trend of each component in the temporal environmental feature sequence is calculated to generate a trend feature vector. The brightness sensitivity component, the visual comfort component, and the trend feature vector are dynamically assembled and sorted according to the regional visual weight to construct a perceptual demand feature set, so that the sensitivity feature weight of the first type of region is higher than that of the second type of region, thereby generating a perceptual demand feature set that matches the current viewing scene.

3. The method according to claim 2, characterized in that, The step of cross-dimensionally associating the temporal environmental feature sequence with the regional content feature matrix, calculating the response correlation degree between the first fluctuation segment and the first type of region, and the adaptation correlation degree between the second fluctuation segment and the second type of region, and generating a cross-sensing correlation matrix includes: The time-domain environmental feature sequence is expanded in dimension, and the fluctuation amplitude feature and duration feature of the first fluctuation segment are normalized and converted into a dimensionless two-dimensional fluctuation feature matrix. Similarly, the features of the second fluctuation segment are converted into a two-dimensional stable feature matrix. The region content feature matrix is ​​standardized in terms of region size and normalized in terms of features. The matrix dimension and numerical range of each display region are unified so that the number of rows, columns and numerical scale of the feature matrix of the first type region and the second type region are consistent, and a standardized region feature matrix is ​​generated. A fixed calculation window is set on the standardized regional feature matrix, and the window size matches the dimension of the two-dimensional fluctuation feature matrix. The window slides in row priority order, and the association weight values ​​between the normalized regional features in the window and the corresponding elements of the two-dimensional fluctuation feature matrix are queried based on the predefined association rule mapping table. The weight values ​​of all windows corresponding to the first fluctuation segment and the first type of region are dynamically weighted and summed to generate a weighted response correlation degree. The weight values ​​are determined according to the weight level in the region visual weight label. The higher the weight level, the larger the weight value. The cumulative weight values ​​of all windows corresponding to the second fluctuation segment and the second type of region are calculated and accumulated according to the spatial position of the region in the image to generate the cumulative adaptive correlation degree. The weighted response correlation degree and the cumulative adaptation correlation degree are filled into a preset matrix frame according to the row and column correspondence. The first fluctuation segment corresponds to the matrix row, the first type of region corresponds to the matrix column, and the second fluctuation segment and the second type of region are similarly generated to generate a cross-sensing correlation matrix.

4. The method according to claim 2, characterized in that, The process involves querying a preset visual perception threshold library based on the cross-perception association matrix to obtain the brightness and color perception thresholds for each region under current ambient light fluctuations. The dynamic range of each threshold is then adjusted according to the region content feature matrix to generate a dynamic perception threshold sequence, including: The cross-sensory association matrix is ​​subjected to threshold filtering. The matrix elements are compared with a preset association strength threshold, and elements that are greater than the association strength threshold are filtered out. The row index and column index corresponding to each filtered element are recorded to generate a list of strongly associated elements. The row index is associated with the corresponding fluctuation segment in the time domain environmental feature sequence, and the column index is associated with the corresponding display area in the regional content feature matrix. Based on the row and column indices in the strongly correlated element list, the corresponding environmental fluctuation segment features and regional content features are extracted from the temporal environmental feature sequence and the regional content feature matrix, and combined into multiple feature query pairs. Based on each set of feature query pairs, the preset visual perception threshold library is queried to obtain the corresponding standard brightness perception threshold and standard color perception threshold, and multiple sets of basic thresholds are generated. The multiple sets of basic thresholds are fused together. For each display area, all the basic thresholds associated with it are weighted and averaged according to the correlation values ​​of the corresponding strongly correlated elements to generate the initial fused brightness threshold and the initial fused color threshold for that area. The brightness distribution density and color transition smoothness of the corresponding region in the region content feature matrix are respectively input into a predefined brightness adjustment mapping function and a color adjustment mapping function. The brightness adjustment mapping function maps the brightness distribution density to a first dimensionless adjustment factor. The brightness adjustment mapping function is configured such that the higher the brightness distribution density, the larger the output first dimensionless adjustment factor. The color adjustment mapping function maps the color transition smoothness to a second dimensionless adjustment factor. The color adjustment mapping function is configured such that the higher the color transition smoothness, the smaller the output second dimensionless adjustment factor. The first dimensionless adjustment factor is used as the brightness threshold adjustment coefficient, and the second dimensionless adjustment factor is used as the color threshold adjustment coefficient. The dynamic range of the initial fused brightness threshold and the initial fused color threshold are scaled by the brightness threshold adjustment coefficient and the color threshold adjustment coefficient, respectively, to generate the final brightness perception threshold and the final color perception threshold for each region. Arrange the final brightness perception threshold and final color perception threshold of all regions in the order of regions to generate a dynamic perception threshold sequence.

5. The method according to claim 1, characterized in that, The step of constructing a brightness mapping relationship by invoking a dynamic mapping mechanism based on the perceived requirement feature set, and generating a dynamic brightness mapping curve that reflects the synergistic effect of ambient light and image content, includes: The set of perceived demand features is layered according to the spatial location of the display area, into a central area feature layer, a transition area feature layer, and an edge area feature layer. Each layer contains the brightness sensitivity index and visual comfort preference index of the corresponding area. Temporal sampling is performed on the central region feature layer, the transition region feature layer and the edge region feature layer respectively. The feature values ​​of each layer at different times are extracted at preset time intervals to generate the temporal feature sequence of each layer. The association strength between the central region feature layer and the transition region feature layer, and the association strength between the transition region feature layer and the edge region feature layer are calculated. The association strength is represented by the ratio of the sum of the product of feature values ​​to the sum of the squares of feature values, and an inter-layer association strength value is generated. Based on the temporal feature sequences of each layer, the corresponding brightness mapping base curves are generated. The base curve of the central region is dominated by the brightness sensitivity index, the base curve of the edge region is dominated by the visual comfort preference index, and the base curve of the transition region combines the two. The brightness mapping base curves are fused based on the inter-layer correlation strength value. The higher the correlation strength value, the greater the weight of the base curves of the adjacent layers. The three base curves are fused into a preliminary brightness mapping curve by weighted averaging. The initial brightness mapping curve is optimized for temporal smoothing. The slope change of the curve at adjacent sampling times is calculated. If the change exceeds a preset threshold, the curve value at the next time step is adjusted to generate a dynamic brightness mapping curve.

6. The method according to claim 5, characterized in that, The calculation of the association strength between the central region feature layer and the transition region feature layer, and the association strength between the transition region feature layer and the edge region feature layer, is expressed by the ratio of the sum of the eigenvalue products divided by the sum of the squares of the eigenvalues, generating interlayer association strength values, including: The temporal feature sequences of the central region feature layer, the transition region feature layer and the edge region feature layer are length-aligned, and the overlapping timestamps in each sequence are truncated so that each sequence contains the same number of feature value samples. The brightness sensitivity index and visual comfort preference index in the time domain feature sequences of each layer are converted into standardized feature values. The feature values ​​are then divided by the standard deviation after subtracting the mean from the feature values ​​to ensure that each feature value is within the same numerical range. Construct the covariance matrix between the central region and the transition region, and the covariance matrix between the transition region and the edge region. The matrix elements are the covariance of the corresponding feature value sequences. The rows represent the features of the central region or the transition region, and the columns represent the features of the transition region or the edge region. The trace of the covariance matrix is ​​calculated and used as the numerator for calculating the correlation strength. The larger the trace value, the larger the sum of the covariances of the two layers of features. Calculate the sum of squares of the eigenvalues ​​of the feature layers in the central region and the feature layers in the transition region. Add the two together and use them as the denominator for calculating the correlation strength. The larger the sum of squares, the greater the total energy of the two feature layers. Divide the numerator by the denominator to obtain the interlayer correlation strength value, and generate the central transition correlation strength value and the transition edge correlation strength value; The step of performing temporal smoothing optimization on the initial brightness mapping curve, calculating the slope change of the curve at adjacent sampling times, and adjusting the curve value at the next time step if the change exceeds a preset threshold, to generate a dynamic brightness mapping curve, includes: The slope of the preliminary brightness mapping curve is calculated by dividing the difference between the curve values ​​at adjacent sampling times by the time interval, and a slope sequence is generated. The slope sequence is subjected to change extraction, the difference between adjacent slope values ​​is calculated, and a slope change sequence is generated. A positive change indicates that the slope is increasing, and a negative change indicates that the slope is decreasing. The threshold range of slope change is determined based on the maximum and minimum values ​​of the preliminary brightness mapping curve. The larger the difference between the maximum and minimum values, the wider the threshold range, thus generating a dynamic upper threshold and a dynamic lower threshold. Traverse the slope change sequence, identify the threshold points where the change exceeds the upper limit of the dynamic threshold or falls below the lower limit of the dynamic threshold, and record the position index of the threshold points and the corresponding slope change. The curve value at the next sampling time corresponding to the threshold point is adjusted and calculated. The curve value is adjusted proportionally according to the deviation between the slope change of the threshold point and the threshold. The larger the deviation, the larger the adjustment. The slope sequence and slope change sequence of the adjusted initial brightness mapping curve are recalculated to verify whether all changes are within the dynamic threshold range. If there are points exceeding the threshold, the threshold point identification and curve value adjustment operations are performed again until all changes meet the smoothness requirements, and a dynamic brightness mapping curve is generated.

7. The method according to claim 1, characterized in that, The invocation of the pre-built partitioned lighting effect model includes: The pre-stored spatial distribution matrix is ​​invoked. The spatial distribution matrix is ​​generated based on the spatial coordinate marking of the MiniLED backlight zoned LED cluster before leaving the factory. It records the two-dimensional coordinates of all LEDs in each zoned LED sub-cluster, and the matrix elements are the set of LED coordinates of the corresponding zone. The pre-generated temperature sensitivity coefficient sequence is invoked. This temperature sensitivity coefficient sequence is generated based on the temperature sensitivity test data of each partition LED sub-cluster before leaving the factory, and describes the variation law of luminous efficacy output under different ambient temperatures. Call the pre-built partition crosstalk matrix, which is generated based on the optical crosstalk test data of adjacent partition LED sub-clusters before leaving the factory, and includes the crosstalk influence between adjacent partitions; The dynamic brightness mapping curve is sampled for regional brightness requirements to generate a target brightness requirement sequence for each region. The spatial distribution matrix, the temperature sensitivity coefficient sequence, the partition crosstalk matrix, and the target brightness requirement sequence are correlated and modeled to generate a partitioned luminous efficacy model that includes luminous efficacy output characteristics and crosstalk effects. The parameters of the partitioned light effect model are adapted by querying the corresponding coefficients in the temperature sensitivity coefficient sequence according to the current ambient temperature, and adjusting the temperature influence parameters in the model so that the model output is adapted to the current temperature conditions.

8. The method according to claim 7, characterized in that, The invocation of the pre-generated temperature sensitivity coefficient sequence includes: Read the pre-stored temperature sensitivity coefficient sequence from the display storage unit. The temperature sensitivity coefficient sequence is generated by fitting the light effect of each zone LED sub-cluster under different ambient temperatures before leaving the factory. It contains sensitivity coefficients corresponding to multiple temperature nodes. The system acquires the real-time temperature value collected by the ambient temperature sensor and matches the real-time temperature value with the temperature nodes in the temperature sensitivity coefficient sequence. If the real-time temperature value matches a certain temperature node, the sensitivity coefficient corresponding to that node is directly called. If the real-time temperature value is between two adjacent temperature nodes, the sensitivity coefficient corresponding to the real-time temperature is calculated by linear interpolation, with the interpolation weight being inversely proportional to the distance between the two nodes. The sensitivity coefficients obtained by matching or interpolation are arranged in partition order to generate a partition temperature sensitivity coefficient vector at the current temperature, which serves as the real-time call result of the temperature sensitivity coefficient sequence. Based on the temperature sensitivity coefficient vector of the partition, the temperature influence term in the partition luminous efficacy model is adjusted so that the luminous efficacy output calculated by the partition luminous efficacy model reflects the actual characteristics at the current temperature. The invocation of the pre-built partition crosstalk matrix includes: Read the pre-stored partition crosstalk matrix from the display storage unit. The partition crosstalk matrix is ​​calculated and generated before leaving the factory by controlling the change of driving conditions of a single partition and collecting the change of light effect of adjacent partitions. The rows represent the interfering partitions, the columns represent the interfering partitions, and the elements are the crosstalk influence degree. Based on the current partition working status of the display screen, identify the set of partitions that are in the active state, set the crosstalk influence of inactive partitions to zero, and generate a crosstalk submatrix for active partitions. The active partition crosstalk submatrix is ​​sparsified by retaining elements whose absolute value is greater than a preset crosstalk threshold and setting the remaining elements to zero to generate a sparse crosstalk matrix. The row and column indices of the sparse crosstalk matrix are mapped to the current partition number so that the partition positions corresponding to the matrix elements are consistent with the actual display screen partition layout, thereby generating a partition crosstalk matrix after position calibration. The impact of changes in the driving conditions of adjacent partitions on the light output of the current partition is calculated based on the partition crosstalk matrix after the position calibration.

9. A brightness control device for a minimized display screen, characterized in that, include: The information acquisition module is used to acquire information on the dynamic changes of ambient light in the current environment and the content feature information of the image currently displayed on the screen. The information on the dynamic changes of ambient light includes the distribution of light intensity at different times, and the content feature information includes the brightness distribution and color distribution of each display area in the image. The demand analysis module is used to input the ambient light dynamic change information and the content feature information into a preset human eye visual perception model to analyze the perception demand and obtain a set of perception demand features that matches the current viewing scene. The set of perception demand features includes the human eye's brightness sensitivity index and visual comfort preference index for different display areas. The relationship construction module is used to call the dynamic mapping mechanism to construct the brightness mapping relationship based on the perception requirement feature set, and generate a dynamic brightness mapping curve that reflects the synergistic effect between ambient light and image content. The dynamic brightness mapping curve is used to describe the nonlinear transformation relationship from input brightness value to output brightness value. The model calling module is used to call the pre-built partitioned light effect model. The partitioned light effect model includes the light effect output characteristics of each partition LED under different driving conditions and the light crosstalk influence between adjacent partitions. The partitioned light effect model is pre-built based on temperature sensitivity test data and light crosstalk test data before leaving the factory. The drive adjustment module is used to perform inverse solving based on the partition light effect model and the dynamic brightness mapping curve to generate partition calibration drive parameters, and send the partition calibration drive parameters to the drive control unit of the MiniLED backlight module to drive brightness adjustment.

10. A display screen, comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 8.

Images