Anti-dazzle Micro LED display screen and anti-dazzle adjusting method and system thereof

Abstract

The invention relates to the technical field of display screens, and discloses an anti-dazzle Micro LED display screen and an anti-dazzle adjusting method and system.The method comprises the steps that multi-dimensional spectral data collection is conducted on incident light, and a multi-dimensional spectral feature matrix is obtained; performing space-time light field reconstruction of tensor decomposition and spherical harmonic function expansion operation on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data; inputting the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; performing anti-dazzle display parameter optimization based on the light field distribution prediction data to obtain a display parameter target value; and executing a hierarchical control strategy according to the display parameter target value, and generating a multi-parameter cooperative control signal for light transmittance adjustment, polarization state modulation and microstructure actuation, the method realizes accurate conversion from the light field characteristics to the control parameters, ensures the stability and continuity of the display effect, effectively reduces the system energy consumption, and improves the operation efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of display screens, and in particular to an anti-glare Micro LED display screen and an anti-glare adjustment method and system thereof. Background Art

[0002] As display devices are increasingly used outdoors and in high-illuminance environments, the importance of anti-glare display technology has become increasingly prominent. Traditional anti-glare technology mainly relies on fixed optical structures or simple feedback adjustment mechanisms, which are difficult to cope with complex and changing light environments, resulting in unstable display effects and poor user experience.

[0003] There are three main technical bottlenecks in the existing anti-glare adjustment methods: the collection of ambient light characteristics is too simplistic, often only considering a single parameter of light intensity, ignoring the multi-dimensional characteristics of light, resulting in insufficient perception of complex light environments; secondly, traditional control strategies lack the ability to deeply analyze and predict the temporal and spatial variation characteristics of the light field, and cannot achieve rapid response and forward-looking adjustment to dynamic light environments; thirdly, single-parameter or simple superposition adjustment methods make it difficult to achieve coordinated optimization of multiple anti-glare parameters, limiting the further improvement of the anti-glare effect. Summary of the invention

[0004] The present invention provides an anti-glare Micro LED display screen and an anti-glare adjustment method and system thereof. The present invention realizes accurate conversion from light field characteristics to control parameters while ensuring the stability and continuity of the display effect, effectively reducing system energy consumption and improving operating efficiency.

[0005] In a first aspect, the present invention provides an anti-glare Micro LED display and an anti-glare adjustment method thereof, wherein the anti-glare Micro LED display and the anti-glare adjustment method thereof include: Collect multi-dimensional spectral data of incident light to obtain a multi-dimensional spectral feature matrix including light intensity value, color temperature value, polarization value and incident angle value; Reconstructing the space-time light field by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors; Inputting the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; Based on the light field distribution prediction data, an anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure is constructed, and the display parameter target value is obtained by solving the model through a multi-objective optimization algorithm; A hierarchical control strategy is executed according to the display parameter target value to generate a multi-parameter coordinated control signal for transmittance adjustment, polarization state modulation and microstructure actuation.

[0006] In a second aspect, the present invention provides an anti-glare Micro LED display and an anti-glare adjustment system thereof, wherein the anti-glare Micro LED display and the anti-glare adjustment system thereof include: The acquisition module is used to collect multi-dimensional spectral data of the incident light to obtain a multi-dimensional spectral feature matrix including light intensity value, color temperature value, polarization value and incident angle value; A light field reconstruction module, used for reconstructing the spatiotemporal light field by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors; A dynamic modeling module, used for inputting the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; A parameter optimization module, for constructing an anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure based on the light field distribution prediction data, and obtaining a display parameter target value by solving the model through a multi-objective optimization algorithm; A generation module is used to execute a hierarchical control strategy according to the display parameter target value to generate a multi-parameter coordinated control signal for transmittance adjustment, polarization state modulation and microstructure actuation.

[0007] In the technical solution provided by the present invention, full-dimensional acquisition of light intensity, color temperature, polarization degree and incident angle is realized through an array spectral sensor, high-precision quantization and synchronous sampling technology are used to ensure the accuracy and real-time performance of data acquisition, and a standardized multi-dimensional spectral feature matrix is ​​constructed to provide a comprehensive data basis for subsequent analysis; tensor decomposition and spherical harmonic function expansion are combined to realize accurate mathematical description of the light field, key features of spatial distribution are obtained through high-order geometric feature extraction, and wavelet multi-resolution analysis is used to capture the change law of the time dimension; a multi-scale convolution structure is used to enhance the feature extraction capability, an attention mechanism is introduced to improve the feature fusion effect, and forward-looking prediction is realized through an encoder-decoder structure; an accurate mapping model of display parameters is established to realize accurate conversion from light field features to control parameters, a multi-objective optimization algorithm is used to balance display clarity, color reproduction and energy efficiency, and a hierarchical control strategy is used to realize the coordinated adjustment of transmittance, polarization angle and microstructure; in addition, the present invention realizes multi-level response speed to meet the adjustment requirements of different parameters, while ensuring the stability and continuity of the display effect, effectively reducing system energy consumption and improving operation efficiency.

[0008] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0009] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of an embodiment of an anti-glare Micro LED display and an anti-glare adjustment method thereof in an embodiment of the present invention; Figure 2 Schematic diagram of an embodiment of an anti-glare Micro LED display and an anti-glare adjustment system thereof in an embodiment of the present invention. DETAILED DESCRIPTION

[0011] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0012] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.

[0013] To facilitate understanding of this embodiment, firstly, an anti-glare MicroLED display screen and an anti-glare adjustment method thereof disclosed in an embodiment of the present invention are described in detail. Figure 1 As shown, the method comprises the following steps: 101. Collect multi-dimensional spectral data of incident light to obtain a multi-dimensional spectral feature matrix including light intensity value, color temperature value, polarization value and incident angle value; It is understandable that the execution subject of the present invention can be an anti-glare Micro LED display and its anti-glare adjustment system, or a terminal or a server, which is not limited here. The embodiment of the present invention is described by taking the server as the execution subject as an example.

[0014] Specifically, the incident light is sampled with high precision using array-distributed spectral sensor units, and the sensor units are ensured to cover the visible light band of 350-780nm, so as to ensure that the acquired spectral data reflects the optical characteristics of the ambient light. The array-distributed method helps to improve the accuracy of spatial sampling, so that the collection of spectral data covers a wider range, while improving the ability to capture incident light at different angles, and enhancing the accuracy of light field reconstruction through the high-density arrangement of sensor units. The original spectral data stream is framed at a high frequency of 1000Hz, that is, the collected spectral data is framed at fixed time intervals to ensure the time consistency of the data, and timestamp information is added to each frame of data, so that the subsequent data processing link can dynamically analyze the spectral data according to the time sequence to obtain a spectral data frame containing timestamp information. Each frame of spectral data is subjected to 16-bit quantized digital conversion processing, and the analog spectral signal is converted into standardized digital spectral data through high-precision analog-to-digital conversion (ADC) technology, and the conversion accuracy is ensured to be high enough to retain the information of the original spectral data to the greatest extent. In this process, the 16-bit quantization accuracy means that the system will characterize the subtle changes in light intensity with a resolution of 65536 levels. The standardized digital spectral data is subjected to feature analysis, including the extraction of components of light intensity, color temperature, polarization degree and incident angle. Among them, the calculation of light intensity is obtained by integrating the spectral data in different wavelength ranges, while the color temperature is extracted from the spectral data by fitting the Plank blackbody radiation curve or using the CIE chromaticity calculation method; the calculation of polarization degree depends on the measurement of polarized light components of a specific wavelength, and the polarization characteristics of light are quantified by calculating the Stokes parameters; the incident angle is obtained by performing trigonometric operations on the measurement results of the array spectral sensor in different directions to form a complete multi-dimensional spectral parameter value. The multi-dimensional spectral parameter values ​​are spatially marked and time-synchronized to obtain a spectral parameter set with time and space identifiers. The spatial position marking process involves using the array distribution information of the sensor unit to calibrate the spectral data at different positions and construct an accurate spectral spatial distribution map; the time synchronization process uses a unified clock system to align the timestamps of the data collected by different sensors to ensure that all data are analyzed in a consistent manner in chronological order. The data structure of the spectral parameter set with time and space identification is reorganized to form a multi-dimensional spectral feature matrix. The matrix is ​​arranged according to the spectral parameter values ​​of different dimensions, and the structure of the matrix is ​​ensured to be compatible with the subsequent tensor calculation and light field modeling requirements. The construction of the multi-dimensional spectral feature matrix requires that light intensity, color temperature, polarization degree and incident angle be used as different feature dimensions of the matrix, while ensuring that each element in the matrix corresponds to a specific time and space position.

[0015] 102. Reconstruct the space-time light field by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors; Specifically, a third-order tensor decomposition operation is performed on the multidimensional spectral feature matrix to extract the core structure of the light field data. In the process of tensor decomposition, Tucker decomposition or CP decomposition method is used, wherein Tucker decomposition generates a core tensor and its corresponding factor matrix, so that the original high-dimensional data is reduced in dimension and mapped to a more compact expression space. The core tensor retains the main features of the data, while the factor matrix describes the change pattern of the light field in different dimensions. Based on the core tensor and the factor matrix, a spherical harmonic function coefficient matrix is ​​constructed, and the light field data is decomposed using the spherical harmonic expansion method. Spherical harmonics is a mathematical tool for describing the angle-dependent characteristics in three-dimensional space. The light field is decomposed at an angle, so that parameters such as light intensity, polarization degree, and incident angle are expressed as linear combinations of spherical harmonic basis functions of different orders. By calculating the spherical harmonic expansion coefficients, the distribution characteristics of the light field in different spatial directions are obtained. After the calculation is completed, the spatial gradient of the spherical harmonic basis function expansion is calculated to obtain the distribution gradient matrix of the light intensity in three-dimensional space. The gradient matrix describes the change trend of the light intensity at different spatial positions, especially in scenes with complex lighting environments, and the gradient information is used to predict the change of the light field. The distribution gradient matrix is ​​subjected to high-order geometric feature extraction to analyze the topological features of the light field in three-dimensional space. Geometric features such as curvature, divergence, and curl are calculated, where curvature describes the degree of bending of the light field in different directions, divergence is used to characterize the divergence or convergence characteristics of light, and curl describes the rotation characteristics of the light field in space. At the same time, the factor matrix is ​​subjected to wavelet multi-resolution analysis in the time dimension to extract the change pattern of the light field in the time dimension. Wavelet analysis is a signal processing method that provides both time and frequency information. It decomposes the light field data into different frequency components and extracts the time feature vector containing trend features, periodic features, and random features. Among them, the trend feature reflects the slow change of the light field on a long time scale, which is related to the changing law of sunlight; the periodic feature describes the periodic oscillation of the light field, which is related to the modulation characteristics of the ambient light source or external interference factors; and the random feature reflects the random fluctuations in the light field on a short time scale, such as light intensity fluctuations caused by cloud cover or dynamic object cover. Through this step, the time information of the light field is decomposed into fine granularity, so that the anti-glare system can more accurately predict future light field changes and adjust the optical parameters of the display to adapt to the changing lighting environment. The spatial feature vector and the time feature vector are outer-producted to construct four-dimensional light field feature data. The light field data not only describes the optical distribution in three-dimensional space, but also reflects the evolution of light over time.

[0016] 103. Input the four-dimensional light field feature data into the deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; Specifically, the four-dimensional light field feature data is input into the temporal feature extraction layer of the deep learning prediction network, and feature decomposition is performed in the layer to obtain the multi-scale temporal features of the light field data. The temporal feature extraction layer adopts three parallel causal convolution modules, each of which contains two convolution layers and a LeakyReLU activation function layer. The role of causal convolution is to ensure that the data of the current time step is only affected by the information of the past time step, so as to maintain the temporal causal relationship and avoid the problem of future information leakage. By performing causal convolution calculations in parallel at multiple time scales, the network captures the light field change patterns of short-term dependence and long-term dependence at the same time, thereby improving the modeling ability of dynamic light fields. LeakyReLU is used as an activation function to alleviate the gradient vanishing problem and allow the network to more flexibly handle small changes in light field data during the learning process, so that the feature extraction layer effectively extracts stable and meaningful temporal features from the four-dimensional light field feature data. The multi-scale temporal features are input into the spatial feature extraction layer of the deep learning prediction network, which enhances the spatial information of the light field to improve the accuracy and stability of the prediction. The spatial feature extraction layer consists of two parallel convolution modules, each of which contains three convolution layers and a batch normalization layer. The convolution layer is used to extract local spatial features, while the batch normalization layer can standardize the data to accelerate the training process and stabilize the gradient flow. By adopting a parallel structure, the network can perform calculations at different spatial scales to simultaneously capture the local and global features of the light field, thereby enhancing the ability to express complex light field distributions. The batch normalization layer can reduce the network's sensitivity to initial parameters and improve gradient propagation, thereby increasing the convergence speed of light field prediction and reducing the risk of overfitting. The multi-scale spatial features are input into the feature fusion layer of the deep learning prediction network, which integrates temporal information and spatial information to construct a complete spatiotemporal fusion feature. In this process, the feature fusion layer uses a multi-head self-attention module and a weighted summation module. The multi-head self-attention module is used to calculate the correlation between different time steps and spatial positions, thereby capturing the dependency of light field data globally and improving the network's ability to model complex light field patterns. Since the change of the light field is not only affected by local environmental factors but also by the overall lighting conditions, the self-attention mechanism effectively models the relationship between the light field at different time and space points, thereby improving the accuracy of the prediction. The role of the weighted summation module is to weightedly fuse different spatiotemporal features to extract the most representative feature information and reduce the impact of redundant data, making the final light field prediction more accurate. The spatiotemporal fusion features are input into the light field prediction layer of the deep learning prediction network to generate light field distribution prediction data. The light field prediction layer adopts an encoder-decoder network structure, in which the encoder part compresses the input high-dimensional light field data to extract the potential features of the light field, and the decoder part reconstructs the compressed data to generate light field prediction results.In the encoder, four downsampling convolution blocks are used for feature extraction. Each downsampling convolution block contains a convolution layer with a stride of 2 and a LeakyReLU activation function layer. The convolution layer with a stride of 2 is used to reduce the resolution of the data to reduce the amount of calculation and extract more abstract features, while the LeakyReLU activation function layer helps to keep the information flow and prevent the gradient from disappearing. By gradually reducing the resolution of the data, the encoder extracts high-level semantic information of the light field data layer by layer and removes redundant details to improve the generalization ability of the network. In the decoder part, four upsampling deconvolution blocks are used for data reconstruction. Each upsampling deconvolution block contains a transposed convolution layer and a ReLU activation function layer. The transposed convolution layer is used to map low-dimensional features back to high-dimensional space to restore the complete structure of the light field, while the ReLU activation function layer is used to increase the nonlinear expression ability of the network, thereby improving the accuracy of predicted details. By adopting a symmetrical upsampling-downsampling structure, the network can restore the local details of the light field while ensuring the global characteristics of the light field, so that the final generated light field distribution prediction data can not only maintain the consistency of the overall trend, but also accurately characterize the local change characteristics of the light field.

[0017] 104. Based on the light field distribution prediction data, an anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure is constructed, and the target value of the display parameter is obtained by solving the multi-objective optimization algorithm; Specifically, the light field distribution prediction data is expanded into a three-dimensional tensor to extract key optical features. The light field data consists of multi-dimensional features, including information such as light intensity, polarization state, and incident angle. Through three-dimensional tensor expansion, the light intensity distribution matrix, polarization state distribution matrix, and incident angle distribution matrix are extracted respectively. The bilinear interpolation method is used to construct the light intensity-transmittance mapping function, polarization state-polarization angle mapping function, and incident angle-microstructure mapping function to form a mapping relationship model of display parameters. The role of bilinear interpolation is to smooth the change relationship between different parameters to ensure the continuity and stability of parameter adjustment during the optimization process, and to avoid discontinuous jump problems caused by data discretization. Through this interpolation process, an accurate correspondence between optical properties and display parameters is established, allowing the optimization model to more finely regulate the anti-glare properties. The value range of the transmittance parameter is set to 5%-95% to cover the display regulation requirements from low transmittance to high transmittance; the value range of the polarization angle parameter is set to 0-180 degrees to adapt to optical regulation in different polarization directions; the value range of the surface microstructure parameter is set to 100nm-1000nm to cover the optical effects of nanoscale microstructures, thereby ensuring the best anti-glare performance under different ambient lighting conditions. A multi-objective optimization function is constructed based on the display parameter mapping relationship model to achieve optimal coordinated control of transmittance, polarization angle and surface microstructure. In this process, the multi-objective optimization function simultaneously considers multiple core performance indicators of anti-glare display, including display clarity, color reproduction and energy efficiency. A display clarity function based on the modulation transfer function is defined to measure the resolution performance of the anti-glare Micro LED display under different transmittance and microstructure conditions, ensuring that the displayed content will not be blurred and distorted due to anti-glare adjustment; at the same time, a color reproduction function based on the CIEDE2000 color difference formula is used to calculate the impact of anti-glare adjustment on color accuracy, so as to maintain the naturalness and consistency of the displayed color during the optimization process; an energy efficiency function based on the display power consumption and brightness ratio is defined to reduce power consumption and improve the energy efficiency ratio of the display while ensuring the anti-glare effect. Through the multi-objective optimization function, the multi-objective optimization algorithm is solved to obtain the Pareto optimal solution set. In the optimization solution process, global optimization methods such as genetic algorithm (GA), particle swarm optimization algorithm (PSO) or non-dominated sorting genetic algorithm (NSGA-II) are used to ensure that the optimal parameter combination that meets different optimization objectives is found in the multi-dimensional parameter space. The solution process of the Pareto optimal solution set involves continuously adjusting the transmittance, polarization angle and microstructure parameters so that each solution reaches the local optimum without losing other objectives, and forms a set of optimal solution candidate sets. Sub-goal weights are assigned and solved for the Pareto optimal solution set to ultimately determine the optimal display parameter target value.According to the needs of specific application scenarios, the analytic hierarchy process or entropy weight method is used to perform weighted calculations on display clarity, color reproduction and energy efficiency, and the optimal combination of transmittance, polarization angle and surface microstructure parameters is determined in combination with the constraints of the specific usage environment.

[0018] The three-dimensional tensor expansion operation is performed on the light field distribution prediction data to convert the four-dimensional light field data into a low-dimensional data structure that is easy to process. The light field data is decomposed according to the spatial dimension, polarization characteristics and incident angle information using tensor decomposition technology to obtain the decomposed light field tensor. Based on the decomposed light field tensor, the spatial dimension is integrated by the projection operator to extract the spatial distribution information of the optical parameters. The role of the projection operator is to project the high-dimensional data in a specific direction, thereby eliminating the influence of non-critical dimensions and allowing the data to be represented in two-dimensional or three-dimensional space. Through this method, the light intensity distribution matrix, polarization state distribution matrix and incident angle distribution matrix are obtained respectively. The light intensity distribution matrix is ​​used to describe the energy distribution of the light field at different positions, while the polarization state distribution matrix is ​​used to represent the polarization characteristics of the light field, including linear polarization, circular polarization and elliptical polarization. At the same time, the incident angle distribution matrix is ​​used to characterize the directional characteristics of the light field when it is incident at different angles. A nonlinear mapping function from light intensity to transmittance is constructed based on the light intensity distribution matrix to establish the relationship between light intensity and transmittance. The adjustment of transmittance directly affects the brightness and visibility of the display screen, while the change of light intensity will have a significant impact on the anti-glare effect, so a nonlinear function is used to characterize the relationship between the two. In this process, an exponential function, a logarithmic function or a high-order polynomial fitting method is used to ensure that the mapping function accurately describes the effect of light intensity on transmittance in different areas. By establishing this mapping relationship, the transmittance is dynamically adjusted under different lighting conditions to ensure the best display effect. At the same time, based on the construction of the light intensity-transmittance mapping function, a dynamic mapping function from polarization state to polarization angle is constructed based on the polarization state distribution matrix. Since the directionality of polarized light directly affects the glare suppression effect of the display screen, in the anti-glare adjustment process, the polarization angle of the display screen is adjusted according to the polarization characteristics of the light field to minimize glare interference. In this process, a dynamic mapping function based on a neural network or an interpolation method is used to ensure that the mapping relationship between the polarization state and the polarization angle can adapt to different optical environments. By introducing a dynamic mapping method, the adjustment of the polarization angle is made more adaptive, thereby improving the stability of the anti-glare effect. At the same time, based on the incident angle distribution matrix, a mapping function from incident angle to microstructure period is constructed to optimize the adjustment strategy of surface microstructure. The periodicity of the surface microstructure directly determines the diffraction characteristics of light, thereby affecting the scattering and reflection characteristics of light. In the anti-glare optimization process, the period of the microstructure is adjusted according to the angle of the incident light to ensure that the light can be redistributed in the appropriate direction and reduce the glare caused by specular reflection. When constructing the mapping function, the physical optics simulation method is used to calculate the microstructure's ability to regulate light at different incident angles, and optimization fitting is performed based on experimental data to ensure that the microstructure period can achieve the optimal anti-glare effect at different angles.The light intensity-transmittance mapping function, the polarization state-polarization angle mapping function and the incident angle-microstructure mapping function are combined to construct a parameter optimization matrix, and finally a display parameter mapping relationship model is obtained.

[0019] 105. A hierarchical control strategy is implemented according to the target values ​​of display parameters to generate multi-parameter coordinated control signals for transmittance adjustment, polarization state modulation and microstructure actuation.

[0020] Specifically, the display parameter target values ​​are decoupled to obtain the transmittance control target value, polarization angle control target value and microstructure control target value respectively, so that each control variable can respond independently and adapt to changes in different time scales. On this basis, the transmittance response model, polarization angle response model and microstructure response model are constructed to describe the dynamic change law of each control variable under different optical environments, and ensure that the system is adaptively adjusted according to the changes in the ambient light field. According to the transmittance response model, polarization angle response model and microstructure response model, a three-layer hierarchical control structure is set to perform anti-glare adjustment according to different time scales, in which the transmittance control is configured as a fast response layer, the polarization angle control is configured as a medium-speed response layer, and the microstructure control is configured as a slow response layer. Since the adjustment of the transmittance depends on the arrangement of liquid crystal molecules or the electronic adjustment of the variable light-transmitting film, its response speed is fast, and it is placed in the fast response layer so that the system can complete the brightness adjustment within milliseconds to adapt to sudden changes in illumination. Polarization angle control involves dynamic modulation of polarization state, which depends on liquid crystal polarization optical elements or variable polarization filters. Its adjustment process is slightly slower than transmittance, but it still needs to be adjusted in a short time, which is classified as a medium-speed response layer. Microstructure control relies on dynamic adjustment of surface microstructures at the nanoscale, such as MEMS micromirror arrays or deformable gratings. Its adjustment process is relatively slow and is configured as a slow response layer to optimize anti-glare properties over a longer time scale. After determining the hierarchical control structure, the corresponding control algorithm is used for different control target values ​​to ensure the stability and accuracy of anti-glare adjustment. Among them, the forward PI-backward PD composite control is used for the transmittance control target value to achieve fast response and high-precision adjustment at the same time. The forward PI control part is used to process the steady-state error of the system and ensure the accuracy of transmittance adjustment, while the backward PD control part is used to improve the dynamic response capability of the system to quickly adjust the transmittance when the ambient light changes, thereby avoiding visual discomfort caused by sudden changes in brightness. At the same time, model predictive control (MPC) is used for the polarization angle control target value. By establishing a system dynamic model, the future state is predicted and the control input is optimized to ensure that the polarization state adjustment can adapt to complex lighting conditions. The advantage of the MPC control method is that it can simultaneously consider the dynamic characteristics and constraints of the system, thereby achieving the optimal polarization angle adjustment and reducing the impact of glare. For the microstructure control target value, self-disturbance rejection control is used to improve the system's anti-disturbance ability by observing external interference and adjusting the control strategy in real time. The core of self-disturbance rejection control is to use an extended state observer to estimate the system's disturbance in real time, and to perform adaptive adjustment through nonlinear state error feedback, so that the microstructure actuation remains stable under different light field environments, improving the long-term adaptability of the anti-glare effect. The transmittance adjustment signal, polarization state modulation signal and microstructure actuation signal are synchronously scheduled to ensure the coordinated work of each control variable, and finally generate a multi-parameter collaborative control signal.Consider time alignment and mutual constraints during signal scheduling to avoid interference between different control variables. For example, in a high light intensity environment, the adjustment of transmittance should be performed first, while the adjustment of polarization angle should be adapted according to the change of transmittance, and the adjustment of microstructure should be optimized based on long-term trends to ensure the coordination and stability of the overall anti-glare adjustment. Through the implementation of hierarchical control strategies, precise adjustment of transmittance, polarization state and microstructure is achieved, so that the anti-glare Micro LED display can be dynamically optimized according to the changes in the ambient light field to improve display quality and visual comfort.

[0021] In the embodiment of the present invention, an array spectral sensor is used to realize full-dimensional acquisition of light intensity, color temperature, polarization degree and incident angle, high-precision quantization and synchronous sampling technology are used to ensure the accuracy and real-time performance of data acquisition, and a standardized multi-dimensional spectral feature matrix is ​​constructed to provide a comprehensive data basis for subsequent analysis; tensor decomposition and spherical harmonic function expansion are combined to realize accurate mathematical description of the light field, key features of spatial distribution are obtained by high-order geometric feature extraction, and wavelet multi-resolution analysis is used to capture the change law of the time dimension; a multi-scale convolution structure is used to enhance the feature extraction capability, an attention mechanism is introduced to improve the feature fusion effect, and a forward-looking prediction is realized through an encoder-decoder structure; an accurate mapping model of display parameters is established to realize accurate conversion from light field features to control parameters, a multi-objective optimization algorithm is used to balance display clarity, color reproduction and energy efficiency, and a hierarchical control strategy is used to realize the coordinated adjustment of transmittance, polarization angle and microstructure; in addition, the present invention realizes multi-level response speed to meet the adjustment requirements of different parameters, while ensuring the stability and continuity of the display effect, effectively reducing system energy consumption and improving operation efficiency.

[0022] In a specific embodiment, the process of executing step 101 may specifically include the following steps: The incident light is sampled by array-distributed spectral sensor units to obtain a raw spectral data stream covering the 350-780nm band; The original spectral data stream is subjected to data framing processing at a sampling frequency of 1000 Hz to obtain a spectral data frame containing timestamp information; Perform 16-bit quantized digital conversion processing on the spectrum data frame to obtain standardized digital spectrum data; Extract the components of light intensity, color temperature, polarization degree and incident angle from the standardized digital spectrum data to obtain multi-dimensional spectrum parameter values; The multi-dimensional spectral parameter values ​​are spatially marked and time synchronized to obtain a spectral parameter set with spatiotemporal identifiers, and the spectral parameter set with spatiotemporal identifiers is reorganized to obtain a multi-dimensional spectral feature matrix.

[0023] Specifically, the incident light is sampled by array-distributed spectral sensor units to obtain comprehensive spectral information. In this process, the spectral sensor units are evenly distributed on the display surface or at key positions of the optical system to ensure that the incident light data of different angles and intensities can be fully collected. Each spectral sensor unit detects light signals in the 350-780nm band, which covers the visible light range, ensuring that the optical characteristics of the light source can be accurately captured. is the spectral signal, where is the position of the sensor in space, is the wavelength, is a time variable, and the spectral signal collected by each sensor is expressed as: in, represents the light intensity distribution of the incident light, Represents polarization state information. The original spectral data stream is framed according to the sampling frequency of 1000 Hz to generate spectral data frames containing timestamp information. Assume that the spectral sampling period is Seconds, that is, one frame of spectral data is collected every millisecond, and each frame of data contains spectral components at multiple wavelengths. In order to achieve stable data stream transmission, each frame of spectral data is attached with timestamp information To ensure that the subsequent time synchronization processing accurately matches the spectral data of different time periods. The spectral data frame is digitally converted to 16 bits to obtain standardized digital spectral data. In this process, the analog spectral signal After digitization by analog-to-digital conversion (ADC), the conversion process is expressed as: in, is the digitized spectral data, and Respectively represent the maximum and minimum spectral intensity values ​​within the detection range of the system. Through 16-bit quantization, a resolution of 65536 levels is obtained to ensure that the data has sufficient accuracy for subsequent optical analysis. After obtaining standardized digital spectral data, key optical parameters such as light intensity, color temperature, polarization degree and incident angle are extracted. Among them, light intensity Obtained by integrating the spectral data within a specific wavelength range: Color Temperature It is obtained by fitting the spectral data to the Planck blackbody radiation curve. The calculation formula is as follows: in, is the Wien displacement constant, and is the Planck constant related parameter, is the peak wavelength of the spectrum. It is calculated by the Stokes parameter, which is defined as follows: in, They are Stokes parameters, which are used to describe the polarization characteristics of light. The change of light intensity at different angles is calculated: in, and Respectively represent the light intensity along two orthogonal polarization directions. After completing the extraction of optical parameters, the multi-dimensional spectral parameter values ​​are spatially marked and time synchronized to ensure the integrity and accuracy of the data. The spatial position marking is achieved by matching the geometric coordinates of the sensor array. Assuming that the sensor array adopts The distribution structure of each sensor is defined as: in, is the row and column index of the sensor, is the distance between adjacent sensors. The time synchronization process uses a global clock synchronization algorithm to ensure that the data of different sensors are analyzed under the same time reference. Assume that the time offset of a sensor is , the timestamp after synchronization is: Through synchronous processing, all spectral data are calculated under the same time base to improve the accuracy of data analysis. The data structure of the spectral parameter set with time and space identification is reorganized to form a multi-dimensional spectral feature matrix. The construction of this matrix is ​​expressed as: Each element corresponds to spectral feature data at a spatial position and is expanded into a multidimensional data structure including polarization degree and incident angle.

[0024] In a specific embodiment, the process of executing step 102 may specifically include the following steps: Perform a third-order tensor decomposition operation on the multidimensional spectral feature matrix to obtain the core tensor and its corresponding factor matrix; Based on the core tensor and factor matrix, the spherical harmonic function coefficient matrix is ​​constructed to obtain the spherical harmonic basis function expansion of the light field, and the spatial gradient of the spherical harmonic basis function expansion is calculated to obtain the distribution gradient matrix of light intensity in three-dimensional space; Perform high-order geometric feature extraction on the distribution gradient matrix to obtain a spatial feature vector containing curvature, divergence and rotation; Perform wavelet multi-resolution analysis on the factor matrix in the time dimension to obtain the time feature vector containing trend features, period features and random features; The spatial feature vector and the temporal feature vector are outer-producted to obtain four-dimensional light field feature data.

[0025] Specifically, the multidimensional spectral feature matrix is ​​represented as a third-order tensor ,in represents the first direction of the spatial dimension (such as the horizontal coordinate), represents the second direction of the spatial dimension (such as the vertical coordinate), and Represents the index of the spectral or time dimension. In order to reduce the dimension and extract features of this tensor, Tucker decomposition is used: in, is the core tensor, , , are the corresponding factor matrices respectively. Core tensor The main information of the light field data is retained, while the factor matrix It is used to describe the projection of data in different dimensions. The spherical harmonic function coefficient matrix is ​​constructed based on the core tensor and factor matrix to obtain the angular characteristics of the light field. The representation of the light field in the spherical coordinate system is described by the spherical harmonic expansion, let represents the angular distribution function of the light field, where is the polar angle, is the azimuth angle, then the spherical harmonic expansion is: in, is a spherical harmonic function, defined as: in, is the associated Legendre polynomial, and are the coefficients of the spherical harmonic expansion and are calculated by the following integral: in Represents the integral region. Through this expansion, the light field data is projected onto the spherical harmonic basis, and its spatial gradient is calculated to obtain the distribution gradient matrix of the light intensity in three-dimensional space. In order to analyze the spatial characteristics of the light field, high-order geometric features are extracted from the distribution gradient matrix, especially the curvature, divergence and curl. Assume that the gradient of the light field intensity in three-dimensional space is , then the divergence Defined as: It indicates the degree of diffusion of the light field in the local area. Then describes the rotation characteristics of the light field and is calculated as follows: The curvature Then the curvature tensor is calculated as: These high-order geometric features provide structural information of the light field and are used to optimize the anti-glare adjustment strategy. In order to analyze the temporal variation pattern of the light field, the factor matrix Perform wavelet multi-resolution analysis in the time dimension to extract the trend characteristics, periodic characteristics and random characteristics of the light field. Assume that the time signal is , then the wavelet transform is defined as follows: in, is the wavelet basis function, is the scale parameter, is the translation parameter. Through wavelet decomposition at different scales, the long-term trend of the light field is extracted (low frequency part), periodic component (mid-frequency part) and short-term random components (high frequency part), that is: in, Describes the slow changes of the light field over a long period of time, Describe recurring patterns, and describes random noise and short-time fluctuations. In order to construct the complete four-dimensional light field feature data, the spatial feature vector and the temporal feature vector are outer-producted to combine the spatial and temporal information. Assuming that the spatial feature vector consists of divergence, curl, and curvature, while the time eigenvector Composed of trend, cycle and random components, the four-dimensional light field characteristic data The calculation is as follows: in Represents the outer product operation, the result is a fourth-order tensor , describing the changing characteristics of the light field in space and time.

[0026] Among them, high-order geometric features are extracted from the distribution gradient matrix to obtain a spatial feature vector containing curvature, divergence and curl, including: multi-scale feature subspace decomposition of the distribution gradient matrix, and subspace coefficient matrices of different frequency scales are obtained through wavelet transform; a geometric feature optimization function G(x) is constructed based on the subspace coefficient matrix, including the curvature term Ck, the divergence term Dv and the curl term Rt, to obtain a feature constraint function; Lagrange multipliers λ are set for the feature constraint function, and a composite optimization objective function L(x,λ)=G(x)+λH(x) is constructed, where H(x) is the light field geometric consistency constraint, to obtain an optimization problem; the optimization problem is solved by the gradient descent method, the iterative update step size is set to 0.01, the maximum number of iterations is 1000, and the optimal feature coefficients are obtained; subspace reconstruction is performed based on the optimal feature coefficients, and an enhanced distribution gradient field is obtained through inverse wavelet transform; geometric feature calculation is performed on the enhanced distribution gradient field, and curvature, divergence and curl information are extracted to obtain a spatial feature vector.

[0027] In a specific embodiment, the process of executing step 103 may specifically include the following steps: The four-dimensional light field feature data is input into the temporal feature extraction layer of the deep learning prediction network for feature decomposition to obtain multi-scale temporal features. The temporal feature extraction layer has three parallel causal convolution modules, each of which contains two convolution layers and a LeakyReLU activation function layer. The multi-scale temporal features are input into the spatial feature extraction layer of the deep learning prediction network for feature enhancement to obtain multi-scale spatial features. The spatial feature extraction layer contains two parallel convolution modules, each of which contains three convolution layers and one batch normalization layer. The multi-scale spatial features are input into the feature fusion layer of the deep learning prediction network for processing to obtain spatiotemporal fusion features. The feature fusion layer includes a multi-head self-attention module and a weighted summation module. The spatiotemporal fusion features are input into the light field prediction layer of the deep learning prediction network for light field prediction to obtain light field distribution prediction data. The light field prediction layer adopts an encoder-decoder network structure. The encoder contains 4 downsampling convolution blocks, each of which contains a convolution layer with a step size of 2 and a LeakyReLU activation function layer. The decoder contains 4 upsampling deconvolution blocks, each of which contains a transposed convolution layer and a ReLU activation function layer.

[0028] Specifically, define the four-dimensional light field feature data ,in and represents the spatial dimension, Represents the dimension of physical characteristics such as spectrum or polarization, In the time series feature extraction layer, three parallel causal convolution modules are used. Each causal convolution module contains two convolution layers and a LeakyReLU activation function layer. The design of causal convolution is to ensure that there will be no future information leakage when modeling time series. Therefore, for a certain time step , whose output depends only on Assume that the input light field time series data is , then the calculation formula in causal convolution is as follows: in, Indicates The temporal characteristics of the layer, represents the convolution kernel, is the bias term, * represents the convolution operation, and LeakyReLU is used for nonlinear transformation, and its expression is: in is the LeakyRelU slope parameter, and its value is 0.01 to avoid neuron death. Through three parallel causal convolution modules, each module uses convolution kernels of different scales, such as and , thereby extracting feature information at different time scales, so that the network takes into account both short-term dependencies and long-term trends, and generates multi-scale temporal features. The multi-scale temporal features are input into the spatial feature extraction layer of the deep learning prediction network to enhance the spatial information of the light field and obtain multi-scale spatial features. The spatial feature extraction layer consists of two parallel convolution modules, each of which contains three convolution layers and a batch normalization layer to improve computational stability and prevent gradient disappearance. Suppose the input spatial feature is , then the calculation formula in the spatial convolution layer is: Among them, the calculation method of the batch normalization layer BN is as follows: in and are the mean and standard deviation of the small batch data, and It is a trainable parameter used to restore the representation ability of the model. Through normalization operation, training is accelerated and the generalization ability of the model is improved. The multi-scale spatial features are input into the feature fusion layer of the deep learning prediction network to obtain the spatiotemporal fusion features. This layer contains a multi-head self-attention module and a weighted summation module. The calculation method of multi-head self-attention is as follows: in, They represent the query matrix (Query), key matrix (Key) and value matrix (Value) respectively. is the dimension of the key vector. Through the multi-head self-attention mechanism, the network focuses on the interactive relationship between different time steps and spatial positions in the light field data to extract more discriminative features. The weighted summation module is used to weightedly combine the features of different channels to ensure that information from different modalities can work together to optimize the accuracy of light field prediction. The spatiotemporal fusion features are input into the light field prediction layer, and the encoder-decoder network structure is used for light field prediction. The encoder part consists of 4 downsampling convolution blocks, each of which contains a convolution layer with a step size of 2 and a LeakyReLU activation function layer. The calculation process is: Among them, the convolution operation with a step size of 2 effectively reduces the dimension of the data while retaining key global features. The decoder part contains 4 upsampling deconvolution blocks, each of which contains a transposed convolution layer and a RelU activation function layer. The calculation formula is as follows: The transposed convolution operation remaps the low-dimensional features to the original space to restore the complete distribution of the light field. The output of the decoder is the light field distribution prediction data. , reflecting the light field change trend in future time steps, thereby providing an accurate adaptive adjustment basis for anti-glare display.

[0029] Among them, the four-dimensional light field feature data is input into the deep learning prediction network for dynamic modeling, and the light field distribution prediction data is obtained, which also includes the step of delay compensation processing of the light field distribution prediction data: frequency domain transformation of the light field distribution prediction data to obtain the frequency domain feature sequence of the light field distribution; phase correction of the frequency domain feature sequence, and construction of a phase compensation function based on the sensor sampling delay and the network transmission delay to obtain the compensated frequency domain sequence; amplitude analysis is performed based on the compensated frequency domain sequence, and the amplitude is corrected by a delay-related normalization factor to obtain normalized frequency domain data; the normalized frequency domain data is inversely transformed to obtain a time domain light field distribution sequence; signal coupling analysis is performed on the time domain light field distribution sequence, and a coupling matrix model is established to obtain a signal coupling vector; adaptive weight allocation is performed on the signal coupling vector, and the weight coefficient is dynamically adjusted by a recursive least squares algorithm to obtain a compensation weight coefficient; the time domain light field distribution sequence is dynamically compensated based on the compensation weight coefficient to obtain a compensated light field distribution sequence; stability analysis is performed on the compensated light field distribution sequence, and the system stability is verified by a Lyapunov function to obtain the light field distribution prediction data after delay compensation.

[0030] In this embodiment, before constructing the anti-glare display parameter optimization model, the light field distribution prediction data also includes the following steps: performing quality evaluation calculation on the light field distribution prediction data, constructing a display quality evaluation function Q(x, t) through a variable step-size normalized error evaluation algorithm, wherein x represents the spatial coordinate and t represents the time dimension, and obtaining a display quality evaluation index; constructing an adaptive filtering network based on the display quality evaluation index, wherein the input layer of the adaptive filtering network includes three channels of light intensity, color temperature, and polarization degree, and the hidden layer includes N adaptive filtering units, and the step size μ of each filtering unit is dynamically adjusted through a back propagation algorithm to obtain an adaptive filtering parameter; performing display noise modeling on the adaptive filtering parameters, using wavelet decomposition to divide the display noise into a high-frequency component NH and a low-frequency component NL, and establishing a noise suppression function S(NH, NL) to obtain a noise suppression model; performing noise elimination processing on the light field distribution prediction data based on the noise suppression model. The improved Wiener filtering algorithm is used, and the iteration step size is set to 0.01 to obtain the denoised light field data; the denoised light field data is compensated for display quality, and the compensation function C(x, t) is constructed to realize dynamic compensation for light intensity fluctuation, color distortion and polarization state distortion, and the compensated light field data is obtained; the recursive least squares algorithm is used to identify the parameters of the compensated light field data, and a mapping relationship H(p, q) between display parameters and display quality is established, where p represents the display parameter vector and q represents the quality index vector, and a display parameter mapping model is obtained; display quality constraints are constructed based on the display parameter mapping model, and light intensity uniformity, color reproduction and polarization stability are set as hard constraints, and display power consumption is set as soft constraints, and optimization constraints are obtained; the optimization constraints are input into the adaptive weight allocator, and the constraints are dynamically weighted by fuzzy rules to obtain the final anti-glare display parameter optimization target.

[0031] In a specific embodiment, the process of executing step 104 may specifically include the following steps: The light field distribution prediction data is expanded into a three-dimensional tensor to extract the light intensity distribution matrix, polarization state distribution matrix and incident angle distribution matrix. The light intensity-transmittance mapping function, polarization state-polarization angle mapping function and incident angle-microstructure mapping function are constructed by bilinear interpolation to obtain a display parameter mapping relationship model, in which the transmittance parameter ranges from 5% to 95%, the polarization angle parameter ranges from 0 to 180 degrees, and the surface microstructure parameter ranges from 100nm to 1000nm. A multi-objective optimization function is constructed based on the display parameter mapping relationship model. The multi-objective optimization function includes a display clarity function based on the modulation transfer function, a color reproduction function based on CIEDE2000, and an energy efficiency function based on the ratio of display power consumption to brightness. Through the multi-objective optimization function, the multi-objective optimization algorithm is solved to obtain the Pareto optimal solution set, and the sub-objective weights of the Pareto optimal solution set are allocated and solved to obtain the display parameter target value.

[0032] Specifically, the light field distribution prediction data is defined as a third-order tensor ,in and represents the spatial dimension, Represents the optical feature dimensions, including light intensity, polarization state, and incident angle. In order to expand this tensor into different physical property matrices, the tensor decomposition technique is used, namely: in, To decompose the weights, are the mode matrices of light intensity, polarization state and incident angle respectively. Through this decomposition, the light intensity distribution matrix is ​​extracted respectively , polarization state distribution matrix and the incident angle distribution matrix In order to construct the mapping relationship of optical parameters, the bilinear interpolation method is used to map the light field characteristics to the display parameter space. For the mapping from light intensity to transmittance, let transmittance is a function of light intensity, then it is obtained by bilinear interpolation: in, is the interpolation weight, satisfying , and is calculated based on the distance between pixels. Similarly, for the mapping of polarization state to polarization angle, define: in, Indicates the polarization angle, the range is set from 0 to . For the mapping of incident angle to microstructure period, we have: in, Represents the period of the microstructure, and the range is set to 100 to 1000 nm. After obtaining the display parameter mapping relationship model, a multi-objective optimization function is constructed to balance display clarity, color reproduction and energy efficiency. The display clarity objective function is defined based on the modulation transfer function (MTF): in, is the spatial frequency The modulation transfer function under is the cutoff frequency of the display. Secondly, the color reproduction objective function is calculated based on CIEDE2000: in, It is the first The color difference of the color blocks indicates the degree of color shift perceived by the human eye. Define the energy efficiency objective function based on the ratio of display power consumption to brightness: in, Indicates display brightness. Represents input power consumption. After constructing the optimization function, the multi-objective optimization algorithm is used to solve the Pareto optimal solution set. Assume that the optimization objective function is: A set of Pareto optimal solutions is obtained by using the non-dominated sorting genetic algorithm (NSGA-II) or the particle swarm optimization algorithm, namely: Among them, the collection represents the set of optimal solutions under different weight configurations. In order to select the optimal solution, the sub-goal weights are assigned. For example, the weight is : in By adjusting the weight distribution, the optimal anti-glare parameters for different application scenarios are selected. After weight normalization calculation, the optimal display parameter target value is obtained: This optimization process ensures that the anti-glare Micro LED display maintains optimal display performance under different lighting conditions.

[0033] In a specific embodiment, the execution step performs a three-dimensional tensor expansion on the light field distribution prediction data, extracts the light intensity distribution matrix, the polarization state distribution matrix and the incident angle distribution matrix, and constructs a light intensity-transmittance mapping function, a polarization state-polarization angle mapping function and an incident angle-microstructure mapping function through bilinear interpolation to obtain a display parameter mapping relationship model. The process may specifically include the following steps: Perform a three-dimensional tensor expansion operation on the light field distribution prediction data to obtain a decomposed light field tensor; Based on the decomposed light field tensor, the spatial dimension is integrated through the projection operator to obtain the light intensity distribution matrix, polarization state distribution matrix and incident angle distribution matrix; A nonlinear mapping function from light intensity to transmittance is constructed based on the light intensity distribution matrix to obtain a light intensity-transmittance mapping function; A dynamic mapping function from polarization state to polarization angle is constructed based on the polarization state distribution matrix to obtain a polarization state-polarization angle mapping function; A mapping function from incident angle to microstructure period is constructed based on the incident angle distribution matrix to obtain the incident angle-microstructure mapping function; The light intensity-transmittance mapping function, the polarization state-polarization angle mapping function and the incident angle-microstructure mapping function are combined into a parameter optimization matrix to obtain a display parameter mapping relationship model.

[0034] Specifically, the light field distribution prediction data is represented as a three-dimensional tensor ,in and represents the spatial dimension, Represents the dimensions of optical parameters, including light intensity , polarization state and the angle of incidence In order to process the tensor more efficiently, high-order singular value decomposition (HOSVD) is used to expand the three-dimensional tensor, and its mathematical expression is: in, is the core tensor, are the basis matrices of the light field data in the spatial and optical feature dimensions, respectively. By decomposing, the redundancy of the data is reduced and the main optical feature components are extracted. In order to obtain the light intensity distribution matrix, polarization state distribution matrix and incident angle distribution matrix, the projection operator is applied based on the decomposed light field tensor Integrate the spatial dimensions to compress the spatial information of the data and construct a two-dimensional feature matrix. The projection operation is expressed as: in, represents the projection area, represents the light intensity distribution, represents the polarization state distribution, Represents the incident angle distribution. Projection operator The contribution of each spatial dimension to the final parameter value is adjusted through the weight function, so that the integrated matrix can accurately reflect the changes in optical properties in the entire spatial area. A nonlinear mapping function from light intensity to transmittance is constructed based on the light intensity distribution matrix to optimize the light transmittance of the display. Assuming the transmittance is the light intensity A nonlinear function is fitted using an exponential model: in, and are the minimum and maximum thresholds of transmittance, is an empirical fitting parameter. The model can better describe the nonlinear relationship between light intensity and transmittance, so that the transmittance changes faster when the light intensity is low, and the transmittance tends to be stable when the light intensity is high, ensuring a natural transition of the display effect. At the same time, in order to construct a dynamic mapping function from polarization state to polarization angle, the polarization state distribution matrix Calculate the appropriate polarization angle The adjustment of the polarization angle mainly depends on the Stokes parameters , and , and its calculation formula is: This formula is used to calculate the direction of the principal axis of polarized light, where and represents the linearly polarized light component, Represents the circularly polarized light component. Through this mapping relationship, the polarization direction of the display screen is adjusted in real time to minimize the glare effect of ambient light. In order to optimize the impact of the microstructure on the optical properties, a mapping function from the incident angle to the microstructure period is constructed. Assuming the microstructure period Depends on the angle of incidence , using the sinusoidal model for fitting: in, nm and nm, respectively represent the minimum and maximum periods of the microstructure. This mapping relationship ensures that the microstructure has a more obvious diffraction effect at a larger incident angle, thereby effectively adjusting the scattering direction of light. The light intensity-transmittance mapping function, polarization state-polarization angle mapping function and incident angle-microstructure mapping function are combined into a parameter optimization matrix to construct a complete display parameter mapping relationship model. Assume that the optimization matrix It is composed of the outputs of the above three mapping functions, and its expression is: in, Represents each spatial point The optimization matrix is ​​used to guide the anti-glare control strategy of the display. For example, when a high-brightness light source (such as sunlight) is directly illuminated, the mapping relationship model automatically calculates the appropriate transmittance to reduce light reflection and improve the visibility of the display, and adjusts the polarization angle to minimize glare. At the same time, the microstructure period is optimized to adjust the scattering path of light, thereby further enhancing the anti-glare effect.

[0035] In a specific embodiment, the process of executing step 106 may specifically include the following steps: Decouple the display parameter target value into the transmittance control target value, polarization angle control target value and microstructure control target value, and construct the transmittance response model, polarization angle response model and microstructure response model; A three-layer hierarchical control structure is set according to the transmittance response model, the polarization angle response model and the microstructure response model, wherein the transmittance control is configured as a fast response layer, the polarization angle control is configured as a medium response layer, and the microstructure control is configured as a slow response layer; The transmittance control target value is subjected to forward PI-backward PD composite control to obtain a transmittance adjustment signal; the polarization angle control target value is subjected to model predictive control to obtain a polarization state modulation signal; the microstructure control target value is subjected to self-disturbance rejection control to obtain a microstructure actuation signal; The transmittance adjustment signal, polarization state modulation signal and microstructure actuation signal are controlled and scheduled synchronously to obtain a multi-parameter coordinated control signal.

[0036] Specifically, the display parameter target value is decoupled into the transmittance control target value, the polarization angle control target value, and the microstructure control target value, so as to construct the transmittance response model, the polarization angle response model, and the microstructure response model respectively. Suppose the display parameter target value is , which includes the transmittance , polarization angle and microstructural parameters , then the decoupling process is expressed as: in, Represents the transmittance control target value, ranging from 5% to 95%, used to adjust the brightness of the display; Represents the target value of polarization angle control, the value range is 0 To 180°, used to adjust the rotation direction of the anti-glare polarizer; The representative microstructure control target value ranges from 100 nm to 1000 nm, which is used to control the periodic morphology of the micro-nanostructure surface to optimize the anti-glare properties. After decoupling, the transmittance response model, polarization angle response model and microstructure response model are constructed to describe the dynamic characteristics of the system. The transmittance response model is represented by a first-order inertial system, that is: in, is the current transmittance, is the time constant of the transmittance system, is the transmittance control gain, is the transmittance control input signal. Similarly, the polarization angle response model is expressed as a second-order system: in, is the current polarization angle, is the damping coefficient, is the natural frequency of the system, Control gain for polarization angle, is the polarization angle control input signal. The microstructure control response model is described as a second-order system affected by disturbance: in, is the current microstructure parameter, is an external environmental interference term, which is used to describe the influence of ambient light and material property changes on the microstructure response. After constructing the response model, a three-layer hierarchical control structure is set to ensure the effectiveness and stability of the control strategy. Among them, transmittance control is configured as a fast response layer, polarization angle control is configured as a medium-speed response layer, and microstructure control is configured as a slow response layer, because the adjustment of transmittance usually depends on liquid crystal or variable light-transmitting film, which has a short response time, usually in milliseconds; while the adjustment of polarization angle involves liquid crystal polarization rotation, which requires tens to hundreds of milliseconds of adjustment time; microstructure control involves nanoscale MEMS or deformable gratings, which have a slow adjustment speed and take several seconds to complete the adjustment. A hierarchical control structure is used to optimize the anti-glare properties at different time scales. In order to achieve precise transmittance control, the transmittance control target value is controlled by combining forward PI (proportional-integral) control with backward PD (proportional-differential) control, that is: in, is the transmittance error, are the gain parameters of the Pl-PD controller respectively. The forward Pl control is used to ensure the steady-state error convergence, and the backward PD control is used to improve the dynamic response speed of the system so that the transmittance can quickly adapt to environmental changes. For polarization angle control, model predictive control is used, and its optimization goal is to minimize the polarization angle error at future moments, that is, in, To predict the time domain length, is the regularization parameter that controls the input, is the optimization variable. By solving the optimization problem, the optimal polarization angle adjustment signal is calculated. For microstructure control, since the system dynamics are affected by external disturbances, the anti-disturbance control method is adopted to improve the anti-disturbance ability of the system. Its control law is: in, is the error signal estimated by the extended state observer, is the estimated disturbance term, is the system parameter. By adaptively adjusting the control input, the influence of external illumination changes on microstructure control is effectively suppressed, and the stability of anti-glare adjustment is improved. The transmittance adjustment signal, polarization state modulation signal and microstructure actuation signal are synchronously scheduled to ensure the coordinated work of each control variable. Assume that the synchronous scheduling strategy adopts the time alignment mechanism, that is: in, They are the update frequencies of transmittance, polarization angle and microstructure control respectively. By selecting a suitable time step, all control signals are calculated within the same time frame, thus achieving multi-parameter coordinated control.

[0037] The above describes the anti-glare Micro LED display and the anti-glare adjustment method thereof in the embodiment of the present invention. The following describes the anti-glare Micro LED display and the anti-glare adjustment system thereof in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, an anti-glare Micro LED display and an anti-glare adjustment system thereof include: The acquisition module 201 is used to collect multi-dimensional spectral data of the incident light to obtain a multi-dimensional spectral feature matrix including light intensity values, color temperature values, polarization degree values ​​and incident angle values; A light field reconstruction module 202 is used to reconstruct the spatiotemporal light field by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors; A dynamic modeling module 203 is used to input the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; The parameter optimization module 204 is used to construct an anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure based on the light field distribution prediction data, and obtain the display parameter target value by solving the multi-objective optimization algorithm; The generation module 205 is used to execute a hierarchical control strategy according to the display parameter target value to generate a multi-parameter coordinated control signal for transmittance adjustment, polarization state modulation and microstructure actuation.

[0038] Through the coordinated cooperation of the above-mentioned components, the full-dimensional acquisition of light intensity, color temperature, polarization degree and incident angle is realized through an array spectral sensor, and high-precision quantization and synchronous sampling technology are used to ensure the accuracy and real-time performance of data acquisition, and a standardized multi-dimensional spectral feature matrix is ​​constructed to provide a comprehensive data basis for subsequent analysis; tensor decomposition and spherical harmonic function expansion are combined to realize the precise mathematical description of the light field, and the key features of the spatial distribution are obtained through high-order geometric feature extraction, and the change law of the time dimension is captured by wavelet multi-resolution analysis; a multi-scale convolution structure is used to enhance the feature extraction capability, an attention mechanism is introduced to improve the feature fusion effect, and a forward-looking prediction is realized through an encoder-decoder structure; an accurate mapping model of display parameters is established to realize the accurate conversion from light field features to control parameters, a multi-objective optimization algorithm is used to balance display clarity, color reproduction and energy efficiency, and a hierarchical control strategy is used to realize the coordinated adjustment of transmittance, polarization angle and microstructure; in addition, the present invention realizes multi-level response speed to meet the adjustment requirements of different parameters, while ensuring the stability and continuity of the display effect, effectively reducing the energy consumption of the system and improving the operating efficiency.

[0039] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0040] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0041] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An anti-glare Micro LED display and an anti-glare adjustment method thereof, characterized in that: The method comprises: Collect multi-dimensional spectral data of incident light to obtain a multi-dimensional spectral feature matrix including light intensity value, color temperature value, polarization value and incident angle value; Reconstructing the space-time light field by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors; Inputting the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; Based on the light field distribution prediction data, an anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure is constructed, and the display parameter target value is obtained by solving the model through a multi-objective optimization algorithm; A hierarchical control strategy is executed according to the display parameter target value to generate a multi-parameter coordinated control signal for transmittance adjustment, polarization state modulation and microstructure actuation.

2. The anti-glare Micro LED display and the anti-glare adjustment method thereof according to claim 1, characterized in that: The multi-dimensional spectral data of the incident light is collected to obtain a multi-dimensional spectral feature matrix including light intensity values, color temperature values, polarization values ​​and incident angle values, including: The incident light is sampled by array-distributed spectral sensor units to obtain a raw spectral data stream covering the 350-780nm band; The original spectral data stream is subjected to data framing processing at a sampling frequency of 1000 Hz to obtain a spectral data frame containing timestamp information; Performing 16-bit quantized digital conversion processing on the spectrum data frame to obtain standardized digital spectrum data; Extracting components of light intensity, color temperature, polarization degree and incident angle from the standardized digital spectrum data to obtain multi-dimensional spectrum parameter values; The multi-dimensional spectral parameter values ​​are subjected to spatial position marking and time synchronization processing to obtain a spectral parameter set with a time-space identifier, and the spectral parameter set with a time-space identifier is reorganized into a data structure to obtain a multi-dimensional spectral feature matrix.

3. The anti-glare Micro LED display and the anti-glare adjustment method thereof according to claim 1, characterized in that: The space-time light field reconstruction by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors includes: Performing a third-order tensor decomposition operation on the multidimensional spectral feature matrix to obtain a core tensor and its corresponding factor matrix; A spherical harmonic function coefficient matrix is ​​constructed based on the core tensor and the factor matrix to obtain a spherical harmonic basis function expansion of the light field, and a spatial gradient calculation is performed on the spherical harmonic basis function expansion to obtain a distribution gradient matrix of light intensity in three-dimensional space; Extracting high-order geometric features from the distribution gradient matrix to obtain a spatial feature vector including curvature, divergence and rotation; Performing wavelet multi-resolution analysis on the factor matrix in time dimension to obtain a time feature vector including trend features, period features and random features; An outer product operation is performed on the spatial feature vector and the temporal feature vector to obtain four-dimensional light field feature data.

4. The anti-glare Micro LED display and the anti-glare adjustment method thereof according to claim 1, characterized in that: The step of inputting the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data includes: Inputting the four-dimensional light field feature data into the temporal feature extraction layer of the deep learning prediction network for feature decomposition to obtain multi-scale temporal features, wherein the temporal feature extraction layer has three parallel causal convolution modules, each of which includes two convolution layers and a LeakyReLU activation function layer; Inputting the multi-scale temporal features into the spatial feature extraction layer of the deep learning prediction network for feature enhancement to obtain multi-scale spatial features, wherein the spatial feature extraction layer comprises two parallel convolution modules, each of which comprises three convolution layers and one batch normalization layer; Inputting the multi-scale spatial features into a feature fusion layer of a deep learning prediction network for processing to obtain spatiotemporal fusion features, wherein the feature fusion layer includes a multi-head self-attention module and a weighted summation module; The spatiotemporal fusion features are input into the light field prediction layer of the deep learning prediction network for light field prediction to obtain light field distribution prediction data, wherein the light field prediction layer adopts an encoder-decoder network structure, the encoder comprises four downsampling convolution blocks, each of which comprises a convolution layer with a step size of 2 and a LeakyReLU activation function layer, and the decoder comprises four upsampling deconvolution blocks, each of which comprises a transposed convolution layer and a ReLU activation function layer.

5. The anti-glare Micro LED display and the anti-glare adjustment method thereof according to claim 1, characterized in that: The anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure is constructed based on the light field distribution prediction data, and the display parameter target value is obtained by solving the multi-objective optimization algorithm, including: Performing a three-dimensional tensor expansion on the light field distribution prediction data, extracting the light intensity distribution matrix, the polarization state distribution matrix and the incident angle distribution matrix, and constructing a light intensity-transmittance mapping function, a polarization state-polarization angle mapping function and an incident angle-microstructure mapping function through bilinear interpolation to obtain a display parameter mapping relationship model, wherein the transmittance parameter value range is 5%-95%, the polarization angle parameter value range is 0-180 degrees, and the surface microstructure parameter value range is 100nm-1000nm; Constructing a multi-objective optimization function based on the display parameter mapping relationship model, wherein the multi-objective optimization function includes a display clarity function based on a modulation transfer function, a color reproduction function based on CIEDE2000, and an energy efficiency function based on a display power consumption and brightness ratio; The multi-objective optimization algorithm is solved through the multi-objective optimization function to obtain a Pareto optimal solution set, and the sub-objective weights are allocated and solved for the Pareto optimal solution set to obtain a display parameter target value.

6. The anti-glare Micro LED display and the anti-glare adjustment method thereof according to claim 5, characterized in that: The three-dimensional tensor expansion is performed on the light field distribution prediction data to extract the light intensity distribution matrix, the polarization state distribution matrix and the incident angle distribution matrix, and the light intensity-transmittance mapping function, the polarization state-polarization angle mapping function and the incident angle-microstructure mapping function are constructed by bilinear interpolation to obtain a display parameter mapping relationship model, including: Performing a three-dimensional tensor expansion operation on the light field distribution prediction data to obtain a decomposed light field tensor; Based on the decomposed light field tensor, integrating the spatial dimensions through a projection operator to obtain a light intensity distribution matrix, a polarization state distribution matrix, and an incident angle distribution matrix; Constructing a nonlinear mapping function from light intensity to transmittance based on the light intensity distribution matrix to obtain a light intensity-transmittance mapping function; Constructing a dynamic mapping function from polarization state to polarization angle based on the polarization state distribution matrix to obtain a polarization state-polarization angle mapping function; Constructing a mapping function from incident angle to microstructure period based on the incident angle distribution matrix to obtain an incident angle-microstructure mapping function; The light intensity-transmittance mapping function, the polarization state-polarization angle mapping function and the incident angle-microstructure mapping function are combined into a parameter optimization matrix to obtain a display parameter mapping relationship model.

7. The anti-glare Micro LED display and the anti-glare adjustment method thereof according to claim 1, characterized in that: The step of executing a hierarchical control strategy according to the display parameter target value to generate a multi-parameter coordinated control signal for transmittance adjustment, polarization state modulation and microstructure actuation includes: Decoupling the display parameter target value into a transmittance control target value, a polarization angle control target value, and a microstructure control target value, and constructing a transmittance response model, a polarization angle response model, and a microstructure response model; A three-layer hierarchical control structure is set according to the transmittance response model, the polarization angle response model and the microstructure response model, wherein the transmittance control is configured as a fast response layer, the polarization angle control is configured as a medium response layer, and the microstructure control is configured as a slow response layer; The transmittance control target value is subjected to forward PI-backward PD composite control to obtain a transmittance adjustment signal; the polarization angle control target value is subjected to model prediction control to obtain a polarization state modulation signal; the microstructure control target value is subjected to self-disturbance rejection control to obtain a microstructure actuation signal; The transmittance adjustment signal, the polarization state modulation signal and the microstructure actuation signal are synchronously scheduled to obtain a multi-parameter coordinated control signal.

8. An anti-glare Micro LED display and an anti-glare adjustment system thereof, characterized in that: The system is used to implement the anti-glare Micro LED display and the anti-glare adjustment method thereof according to any one of claims 1 to 7, the system comprising: The acquisition module is used to collect multi-dimensional spectral data of the incident light to obtain a multi-dimensional spectral feature matrix including light intensity value, color temperature value, polarization value and incident angle value; A light field reconstruction module, used for reconstructing the spatiotemporal light field by performing tensor decomposition and spherical harmonic expansion operations on the multi-dimensional spectral feature matrix to obtain four-dimensional light field feature data including spatial feature vectors and temporal feature vectors; A dynamic modeling module, used for inputting the four-dimensional light field feature data into a deep learning prediction network for dynamic modeling to obtain light field distribution prediction data; A parameter optimization module, for constructing an anti-glare display parameter optimization model including transmittance, polarization angle and surface microstructure based on the light field distribution prediction data, and obtaining a display parameter target value by solving the model through a multi-objective optimization algorithm; A generation module is used to execute a hierarchical control strategy according to the display parameter target value to generate a multi-parameter coordinated control signal for transmittance adjustment, polarization state modulation and microstructure actuation.

Images