Display screen adjusting system based on ambient light

Abstract

The invention relates to the technical field of display screen adjustment, and discloses an ambient light-based display screen adjustment system, which is characterized in that a data acquisition module is used for acquiring initial data of ambient light and acquiring visual feature data of user history records; the classification module is used for classifying the ambient light intensity by adopting a support vector machine method; the clustering module is used for performing clustering analysis on the ambient light color temperature by adopting a clustering algorithm to determine the color temperature type of the ambient light; the integration module is used for integrating results of the classification module and the clustering module to obtain key feature information of ambient light; the calculation module is used for calling the user visual feature data matched with the current ambient light, calculating the optimal adjustment parameters of the display screen and transmitting the optimal adjustment parameters to the control module of the display screen; the control module is used for adjusting the display screen in real time according to the optimal adjusting parameters, and continuously monitoring interaction data between the user and the display screen after the display screen is adjusted; the display effect is improved, and the personalized requirements of users are met.

Description

Technical Field

[0001] The present invention relates to the technical field of display screen adjustment, and in particular to a display screen adjustment system based on ambient light. Background Art

[0002] With the widespread use of electronic devices, display screens, as an important interface for human-computer interaction, have a crucial impact on user experience. Under different ambient light conditions, users have different requirements for display screen brightness, contrast and other parameters. In the existing technology, the display screen adjustment method mostly adopts a fixed adjustment mode, or only uses a simple sensor to obtain the ambient light intensity and then perform linear adjustment. However, this adjustment method has many problems. On the one hand, simple linear adjustment cannot accurately match the user's visual needs in different scenarios. For example, in a strong light environment, even if the display screen brightness is turned up, the displayed content may still be unclear; in a low light environment, excessive brightness will not only be dazzling, but also increase the energy consumption of the device. On the other hand, the fixed adjustment mode lacks consideration of the user's personalized visual habits, making it difficult to meet the usage needs of different users. Summary of the Invention

[0003] The purpose of the present invention is to solve the above problems and to design a display screen adjustment system based on ambient light.

[0004] The present invention provides a display screen adjustment system based on ambient light, the system comprising:

[0005] The data acquisition module is used to collect the initial data of the ambient light in real time using the ambient light sensor and obtain the visual feature data of the user's historical records;

[0006] A classification module is used to classify the ambient light intensity based on the initial data of the ambient light using a support vector machine method;

[0007] A clustering module is used to perform cluster analysis on the color temperature of the ambient light based on the initial data of the ambient light using a clustering algorithm to determine the color temperature type of the ambient light;

[0008] Integration module, used to integrate the results of the classification module and clustering module to obtain the key feature information of the ambient light;

[0009] The calculation module is used to retrieve the user's visual feature data that matches the current ambient light based on the key feature information of the ambient light, build a multimodal fusion model based on Transformer-GNN, optimize and improve the model's adaptive attention mechanism, calculate the comprehensive feature representation under dynamic weight distribution through the model, and calculate the optimal adjustment parameters of the display screen, and transmit the optimal adjustment parameters to the control module of the display screen;

[0010] The control module is used to adjust the display screen in real time according to the optimal adjustment parameters, and continuously monitor the interaction data between the user and the display screen after the display screen adjustment is completed.

[0011] Optionally, in a first implementation of the present invention, the data acquisition module includes:

[0012] Build a submodule to build a federated learning network between multiple distributed devices and a central server to obtain the visual feature data of user history records collected locally by each distributed device;

[0013] The setup submodule is used to set the initial sampling frequency, the initial state vector and the covariance matrix of the Kalman filter. The ambient light sensors of each distributed device collect initial ambient light data according to the initial sampling frequency. The Kalman filter algorithm is used to predict the change trend and predicted value of the ambient light parameters at the next moment through the state transition equation based on the historically collected ambient light data and the state estimation value before the current moment.

[0014] The comparison submodule is used to compare the currently collected ambient light data with the predicted value, calculate the actual change in ambient light, combine the sampling time interval to obtain the rate of change of ambient light, and dynamically adjust the sampling frequency according to the rate of change of ambient light.

[0015] Optionally, in a second implementation of the present invention, the classification module includes:

[0016] The normalization submodule is used to extract features related to the ambient light intensity classification from the initial ambient light data, combine them into a feature vector, perform dimensionality reduction on the feature vector using principal component analysis, and normalize the feature vector after dimensionality reduction;

[0017] The mapping submodule is used to map the normalized data to the high-dimensional Hilbert space through the quantum kernel function, encode it into a quantum state, measure the quantum state, and output the quantum kernel matrix;

[0018] The solution submodule is used to build an SVM model based on the quantum kernel matrix, and use optimization algorithms such as gradient descent to solve the dual problem of SVM to obtain the optimal classification hyperplane;

[0019] The classification submodule is used to classify the ambient light intensity based on the classification hyperplane, determine the ambient light intensity category to which it belongs, and output the classification result.

[0020] Optionally, in a third implementation of the present invention, the clustering module includes:

[0021] The extraction submodule is used to extract color temperature data from the initial data of ambient light and obtain the time series information of color temperature according to the collected timestamp;

[0022] The initialization submodule is used to extract and cluster potential features using the DEC algorithm. The parameters of the GMM are initialized based on the results of the DEC clustering. The time series information is input into the LSTM model. The LSTM model is used to predict the color temperature data. Based on the prediction results and the current color temperature data, the GMM is used to dynamically adjust the cluster center.

[0023] The determination submodule is used to integrate the cluster labels obtained by DEC clustering and the cluster center information adjusted by time-series perception clustering to determine the color temperature type of the ambient light.

[0024] Optionally, in a fourth implementation of the present invention, the initialization submodule includes:

[0025] A design unit is used to design an autoencoder network, which includes an input layer, a hidden layer, and an output layer, input the color temperature data into the autoencoder, and extract the output of the hidden layer as the potential feature of the color temperature data;

[0026] The iterative unit is used to cluster the extracted potential features using the DEC algorithm, initialize the cluster centers, iteratively update the cluster centers and cluster assignments, minimize the clustering loss function until convergence, and thus divide the color temperature data into different categories.

[0027] Optionally, in a fifth implementation of the present invention, the calculation module includes:

[0028] The data preparation submodule is used to extract user visual feature data that matches the current ambient light and convert the key feature information of the ambient light and the user visual feature data into the feature vector form of the graph neural network;

[0029] The fusion submodule is used to perform multimodal fusion network modeling on the output of the data preparation submodule to obtain the fused feature vector;

[0030] The training submodule is used to use the fused feature vector as the state of reinforcement learning, determine the adjustable parameters of the display screen to obtain the action space of reinforcement learning, build the DQN network, and perform reinforcement learning training on the DQN network;

[0031] The optimization submodule is used to dynamically fine-tune the parameters of the reinforcement learning model output by the training submodule, update the Q-value mapping relationship of the model in real time through online learning strategies, and perform incremental optimization of the model based on user interaction data feedback;

[0032] The selection submodule is used to input the fused feature vector into the trained DQN network and select the action with the largest Q value according to the Q value output by the DQN network. The display screen parameters corresponding to this action are the calculated optimal adjustment parameters.

[0033] Optionally, in a sixth implementation of the present invention, the fusion submodule includes:

[0034] A construction unit, configured to use the key feature information of ambient light and the user's visual feature data as nodes in a graph, and to construct edges of the graph based on the association relationship;

[0035] The adjustment unit is used to use the constructed graph as the input of the graph neural network, use historical data to train the graph neural network, and continuously adjust the network parameters to enable the graph neural network to learn the complex correlation pattern between ambient light characteristics and user visual characteristics;

[0036] The first computing unit is used to fuse the key feature information of ambient light and the user's visual feature data through multi-layer propagation and calculation of the graph neural network to obtain a comprehensive feature representation.

[0037] Optionally, in a seventh implementation of the present invention, the DQN network includes an input layer, a hidden layer, and an output layer, wherein the output layer outputs a Q value corresponding to each action.

[0038] Optionally, in an eighth implementation of the present invention, the training submodule includes:

[0039] The selection unit is used to select an action using the DQN network in the current state and apply the action to the display screen;

[0040] A continuous monitoring unit is used to continuously monitor the user's fatigue, use the change in the user's fatigue as a reward signal, and store the current state, action, reward, and next state in the experience replay buffer;

[0041] The second computing unit is used to randomly sample a batch of data from the experience replay buffer for training the DQN network, calculate the loss function of the DQN network based on the sampled data, and use the Adam optimizer to update the parameters of the DQN network to complete the reinforcement learning training of the DQN network.

[0042] Optionally, in a ninth implementation of the present invention, the control module includes:

[0043] A collection submodule is configured to collect, after the display screen is adjusted, user interaction data between the display screen and the user through sensors on the local device, where the interaction data includes at least false touch rate, touch force and frequency, and dwell time on reading text;

[0044] The storage submodule is used to collect the interaction data between the user and the display screen in real time, and to organize and store the interaction data.

[0045] The technical solution provided by the present invention utilizes an ambient light sensor to collect initial ambient light data in real time and obtain visual feature data of user history records; based on the initial ambient light data, a support vector machine method is used to classify the ambient light intensity; based on the initial ambient light data, a clustering algorithm is used to perform cluster analysis on the ambient light color temperature to determine the ambient light color temperature type; the results of the classification module and the clustering module are integrated to obtain key feature information of the ambient light; based on the key feature information of the ambient light, the user's visual feature data that matches the current ambient light is retrieved, and the optimal adjustment parameters of the display are calculated and transmitted to the control module of the display; the display is adjusted in real time according to the optimal adjustment parameters, and after the display adjustment is completed, the interaction data between the user and the display is continuously monitored; the present invention realizes precise adjustment of the display under different ambient light conditions through multi-dimensional analysis of ambient light data and precise matching of user visual features, effectively solving the problem of inaccurate adjustment in the prior art and improving the display effect; fully considering the user's personalized visual habits and physiological characteristics, providing customized display adjustment solutions for different users, meeting the user's personalized needs and improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Various other advantages and benefits will become apparent to those skilled in the art by reading the following detailed description of the preferred embodiment.The accompanying drawings are only for the purpose of illustrating the preferred embodiment and are not to be considered as limiting the present invention.

[0047] Figure 1 A schematic structural diagram of a display screen adjustment system based on ambient light provided by an embodiment of the present invention;

[0048] Figure 2 A schematic diagram of the structure of a data acquisition module provided in an embodiment of the present invention;

[0049] Figure 3 A schematic diagram of the structure of a classification module provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The terms "first," "second," "third," "fourth," and the like (if any) in the description and claims of the present invention and in the accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a particular order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, apparatus, product, or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to these processes, methods, products, or apparatus.

[0051] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 A schematic structural diagram of a display screen adjustment system based on ambient light provided by an embodiment of the present invention, wherein the method specifically comprises the following steps:

[0052] The data acquisition module is used to collect the initial data of the ambient light in real time using the ambient light sensor and obtain the visual feature data of the user's historical records;

[0053] A classification module is used to classify the ambient light intensity based on the initial data of the ambient light using a support vector machine method;

[0054] A clustering module is used to perform cluster analysis on the color temperature of the ambient light based on the initial data of the ambient light using a clustering algorithm to determine the color temperature type of the ambient light;

[0055] Integration module, used to integrate the results of the classification module and clustering module to obtain the key feature information of the ambient light;

[0056] The calculation module is used to retrieve the user's visual feature data that matches the current ambient light based on the key feature information of the ambient light, build a multimodal fusion model based on Transformer-GNN, optimize and improve the model's adaptive attention mechanism, calculate the comprehensive feature representation under dynamic weight distribution through the model, and calculate the optimal adjustment parameters of the display screen, and transmit the optimal adjustment parameters to the control module of the display screen;

[0057] The control module is used to adjust the display screen in real time according to the optimal adjustment parameters, and continuously monitor the interaction data between the user and the display screen after the display screen adjustment is completed.

[0058] In this embodiment, please refer to Figure 2 , the data acquisition module includes:

[0059] Build a submodule to build a federated learning network between multiple distributed devices and a central server to obtain the visual feature data of user history records collected locally by each distributed device;

[0060] The setup submodule is used to set the initial sampling frequency, the initial state vector and the covariance matrix of the Kalman filter. The ambient light sensors of each distributed device collect initial ambient light data according to the initial sampling frequency. The Kalman filter algorithm is used to predict the change trend and predicted value of the ambient light parameters at the next moment through the state transition equation based on the historically collected ambient light data and the state estimation value before the current moment.

[0061] The comparison submodule is used to compare the currently collected ambient light data with the predicted value, calculate the actual change in ambient light, combine the sampling time interval to obtain the rate of change of ambient light, and dynamically adjust the sampling frequency according to the rate of change of ambient light.

[0062] In this embodiment, please refer to Figure 3 , the classification module includes:

[0063] The normalization submodule is used to extract features related to the ambient light intensity classification from the initial ambient light data, combine them into a feature vector, perform dimensionality reduction on the feature vector using principal component analysis, and normalize the feature vector after dimensionality reduction;

[0064] The mapping submodule is used to map the normalized data to the high-dimensional Hilbert space through the quantum kernel function, encode it into a quantum state, measure the quantum state, and output the quantum kernel matrix;

[0065] The solution submodule is used to build an SVM model based on the quantum kernel matrix, and use optimization algorithms such as gradient descent to solve the dual problem of SVM to obtain the optimal classification hyperplane;

[0066] The classification submodule is used to classify the ambient light intensity based on the classification hyperplane, determine the ambient light intensity category to which it belongs, and output the classification result.

[0067] In this embodiment, the clustering module includes:

[0068] The extraction submodule is used to extract color temperature data from the initial data of ambient light and obtain the time series information of color temperature according to the collected timestamp;

[0069] The initialization submodule is used to extract and cluster potential features using the DEC algorithm. The parameters of the GMM are initialized based on the results of the DEC clustering. The time series information is input into the LSTM model. The LSTM model is used to predict the color temperature data. Based on the prediction results and the current color temperature data, the GMM is used to dynamically adjust the cluster center.

[0070] The determination submodule is used to integrate the cluster labels obtained by DEC clustering and the cluster center information adjusted by time-series perception clustering to determine the color temperature type of the ambient light.

[0071] In this embodiment, the initialization submodule includes:

[0072] A design unit is used to design an autoencoder network, which includes an input layer, a hidden layer, and an output layer, input the color temperature data into the autoencoder, and extract the output of the hidden layer as the potential feature of the color temperature data;

[0073] The iterative unit is used to cluster the extracted potential features using the DEC algorithm, initialize the cluster centers, iteratively update the cluster centers and cluster assignments, minimize the clustering loss function until convergence, and thus divide the color temperature data into different categories.

[0074] In this embodiment, the calculation module includes:

[0075] The data preparation submodule is used to extract user visual feature data that matches the current ambient light and convert the key feature information of the ambient light and the user visual feature data into the feature vector form of the graph neural network;

[0076] The fusion submodule is used to perform multimodal fusion network modeling on the output of the data preparation submodule to obtain the fused feature vector;

[0077] The training submodule is used to use the fused feature vector as the state of reinforcement learning, determine the adjustable parameters of the display screen to obtain the action space of reinforcement learning, build the DQN network, and perform reinforcement learning training on the DQN network;

[0078] The optimization submodule is used to dynamically fine-tune the parameters of the reinforcement learning model output by the training submodule, update the Q-value mapping relationship of the model in real time through online learning strategies, and perform incremental optimization of the model based on user interaction data feedback;

[0079] The selection submodule is used to input the fused feature vector into the trained DQN network and select the action with the largest Q value according to the Q value output by the DQN network. The display screen parameters corresponding to this action are the calculated optimal adjustment parameters.

[0080] In this embodiment, the fusion submodule includes:

[0081] A construction unit, configured to use the key feature information of ambient light and the user's visual feature data as nodes in a graph, and to construct edges of the graph based on the association relationship;

[0082] The adjustment unit is used to use the constructed graph as the input of the graph neural network, use historical data to train the graph neural network, and continuously adjust the network parameters to enable the graph neural network to learn the complex correlation pattern between ambient light characteristics and user visual characteristics;

[0083] The first computing unit is used to fuse the key feature information of ambient light and the user's visual feature data through multi-layer propagation and calculation of the graph neural network to obtain a comprehensive feature representation.

[0084] In this embodiment, the DQN network includes an input layer, a hidden layer, and an output layer, wherein the output layer outputs the Q value corresponding to each action.

[0085] In this embodiment, the training submodule includes:

[0086] The selection unit is used to select an action using the DQN network in the current state and apply the action to the display screen;

[0087] A continuous monitoring unit is used to continuously monitor the user's fatigue, use the change in the user's fatigue as a reward signal, and store the current state, action, reward, and next state in the experience replay buffer;

[0088] The second computing unit is used to randomly sample a batch of data from the experience replay buffer for training the DQN network, calculate the loss function of the DQN network based on the sampled data, and use the Adam optimizer to update the parameters of the DQN network to complete the reinforcement learning training of the DQN network.

[0089] In this embodiment, the control module includes:

[0090] A collection submodule is configured to collect, after the display screen is adjusted, user interaction data between the display screen and the user through sensors on the local device, where the interaction data includes at least false touch rate, touch force and frequency, and dwell time on reading text;

[0091] The storage submodule is used to collect the interaction data between the user and the display screen in real time, and to organize and store the interaction data.

[0092] In this embodiment, the ambient light sensor is used to collect initial ambient light data in real time, including but not limited to parameter information such as light intensity and color temperature. At the same time, visual feature data of the user's historical records are obtained from the local database of the user device. This data covers basic information such as the user's age, gender, and vision condition, as well as past operating habits such as manual adjustment of the display screen brightness and contrast, and physiological feature data of viewing the display screen under different ambient light conditions (such as viewing time, blinking frequency, etc.). The collected initial ambient light data and user visual feature data are preliminarily sorted out. The initial sampling frequency, the initial state vector and the covariance matrix of the Kalman filter are set for the adaptive sampling algorithm. The initial sampling frequency is set to a moderate value. To ensure that the ambient light data can be initially obtained; the state vector of the Kalman filter contains the initial estimated values ​​of parameters such as ambient light intensity and color temperature, and the covariance matrix is ​​used to measure the uncertainty of the initial estimate; a federated learning network is built between multiple distributed devices and the central server to establish a secure data communication channel. Each device and the central server agree on the data interaction protocol and encryption method to ensure data transmission security; the local processing module uses the Kalman filter algorithm to predict the change trend and predicted value of parameters such as ambient light intensity and color temperature at the next moment through the state transition equation based on the historically collected ambient light data and the state estimate before the current moment; at the same time, the prediction is updated according to the state transition matrix and the process noise covariance matrix Error covariance matrix; compare the currently collected ambient light data with the predicted value, calculate the actual change in ambient light in terms of intensity, color temperature, etc., and then combine the sampling time interval to obtain the rate of change of ambient light; according to the calculated rate of change of ambient light, dynamically adjust the sampling frequency according to the pre-set rules; if the rate of change of ambient light is large, it means that the ambient light changes drastically, and increase the sampling frequency to capture the change of ambient light more timely; if the rate of change of ambient light is small, reduce the sampling frequency to reduce the amount of data processing and equipment energy consumption; after that, the ambient light sensor continues to collect ambient light data according to the new sampling frequency, and repeats the above process of prediction, calculation of change rate and adjustment of sampling frequency; each distributed device collects ambient light data locally Historical data collected by the ambient light sensor and visual feature data from user history records; for historical ambient light data, the sensor's original measurement value, acquisition time, and other information are extracted; for user visual feature data, the user's age, gender, vision, manual adjustment of display parameters, and physiological characteristics of viewing the display are compiled. Each device performs preliminary processing on the historical ambient light data, extracting feature data related to sensor errors (such as the deviation of multiple measurements in the same environment), encrypting this processed error calibration data and uploading it to the central server. At the same time, the complete visual feature data containing user privacy information is not uploaded, only the processed features used for collaborative calibration are uploaded.The central server receives error calibration data from different devices and aggregates this data using a federated learning aggregation algorithm (such as the federated averaging algorithm) to generate a global sensor error calibration model. This model reflects the comprehensive situation and patterns of sensor errors across multiple devices. The central server distributes the global sensor error calibration model to each distributed device. After receiving the model, each device uses it to calibrate the current data collected by its local ambient light sensor, correcting the sensor measurement error and obtaining more accurate ambient light data. Simultaneously, the calibrated ambient light data is integrated with locally stored visual feature data from user history records to form an effective dataset for subsequent display adjustments.

[0093] In this embodiment, the ambient light sensor is used to obtain the initial data of the ambient light. These data may include multi-dimensional information such as light intensity, color, and angle. At the same time, the user's operation records under different ambient light conditions are collected as auxiliary information. The data are checked for missing values, abnormal values, etc. For missing values, the mean, median, etc. can be used to fill them. For abnormal values, they are corrected or eliminated according to the actual situation. Feature extraction: Features related to the ambient light intensity classification are extracted from the initial ambient light data and combined into feature vectors. Methods such as principal component analysis (PCA) can be used to reduce the dimension of the features to reduce the amount of calculation. A suitable quantum circuit structure is selected and the characteristics of quantum computing are used to calculate. Calculate the kernel function; through the superposition and entanglement characteristics of quantum bits, efficiently calculate the similarity between samples in high-dimensional space; encode the feature vector of the initial ambient light data and map it to the quantum state space; execute the quantum circuit, measure the quantum state, and obtain the calculation result of the kernel function; model training: construct the SVM model based on the kernel matrix calculated by the quantum kernel function; use the ambient light feature vectors in the training data set and their corresponding ambient light intensity classification labels to determine the model parameters (such as Lagrange multipliers, bias terms, etc.) by solving the SVM optimization problem; gradient descent and other optimization algorithms can be used to solve the dual problem of SVM to obtain the optimal classification hyperplane; continuously monitor new ambient light data , when a new lighting scene appears (such as a sudden switch to indoor or outdoor light), the new data is marked and collected; the new data is added to the existing training data set; the SVM model is updated using the incremental learning method without retraining the entire model; an online learning algorithm, such as the incremental version of Sequential Minimization (SMO), can be used to adjust the model parameters according to the new data so that the model can adapt to the new lighting scene; for the current SVM model, an adversarial training algorithm (such as the Fast Gradient Signed Method FGSM) is used to generate adversarial samples; adversarial samples are small perturbations added to the original samples, causing the model to misclassify these samples; the generated adversarial samples should be Simulate data features under extreme lighting conditions (such as strong glare); Adversarial training: add the generated adversarial samples to the training data set; retrain the SVM model so that the model can learn on both normal samples and adversarial samples; by continuously adjusting the model parameters, the model can better resist the interference of adversarial samples and improve its robustness under extreme lighting conditions; Step 5: Ambient light intensity classification feature input: input the feature vector of the preprocessed ambient light initial data into the trained and enhanced SVM model; the model calculates the input feature vector based on the determined classification hyperplane and parameters, determines the ambient light intensity category it belongs to (such as strong light, medium light, weak light, etc.), and outputs the classification result.

[0094] In this embodiment, an ambient light sensor is used to continuously collect ambient light color temperature data and record the acquisition timestamp to obtain color temperature time series information. In addition, historical user preferences for different color temperatures are collected, such as records of users manually adjusting the display color temperature. The collected color temperature data is checked for missing values ​​or outliers. Missing values ​​can be filled using methods such as linear interpolation; outliers can be identified and corrected using statistical methods (such as those based on standard deviation). Normalization processing: The color temperature data is normalized to a range between [0, 1] to eliminate the impact of different data scales on subsequent clustering analysis. User preference data is also normalized accordingly. An autoencoder network is designed, consisting of an input layer, a hidden layer, and an output layer. The input layer receives preprocessed color temperature data, the hidden layer extracts the data's latent features, and the output layer attempts to reconstruct the input data. The collected color temperature data is used to train the autoencoder. The autoencoder parameters are adjusted by minimizing the error (such as mean squared error) between the input data and the reconstructed data, enabling the autoencoder to learn the latent feature representation of the color temperature data. The preprocessed color temperature data is fed into a trained autoencoder, and the output of the hidden layer is extracted as the latent features of the color temperature data. The extracted latent features are clustered using the DEC algorithm. The cluster centers are initialized. Then, by iteratively updating the cluster centers and the cluster assignments of the samples, the clustering loss function is minimized until convergence, thereby classifying the color temperature data into different categories. An LSTM network is designed, which takes as input a sequence of color temperature data with timestamps and outputs predictions of future color temperature data. The LSTM model is trained using historical color temperature time series data. The parameters of the LSTM model are adjusted by minimizing the error between the predicted and actual values, enabling it to learn the temporal variation of color temperature. Based on the results of the DEC clustering, the parameters of the GMM are initialized, including the mean, covariance, and weights of each Gaussian component. The trained LSTM model is used to predict future color temperature data. The GMM dynamically adjusts the cluster centers based on the predicted results and the current color temperature data. Specifically, the GMM parameters are updated based on the new data, so that the cluster centers can adapt to temporal variations in color temperature (such as those caused by sunrise and sunset). Color temperature-related features are extracted from collected historical user preference data, such as the user's preferred color temperature range and preferred color temperature change pattern. A collaborative filtering algorithm is used to recommend color temperature categories that each user might like based on similarities between users. Specifically, similarities between users are calculated (e.g., based on cosine similarity), and other users similar to the target user are identified. Color temperature categories are then recommended based on the preferences of these similar users. Based on the collaborative filtering recommendations, a personalized color temperature preference label is generated for each user, such as "warm color preference user" or "cool color preference user."The cluster labels obtained by DEC clustering, the cluster center information adjusted by time-series-aware clustering, and the preference labels obtained by user-customized clustering are integrated to form the key characteristic information of the ambient light. Based on this integrated key characteristic information, the color temperature type of the ambient light is determined. For example, combining the clustering results with user preferences, it can determine whether the current ambient light color temperature is warm, suitable for users who prefer warm colors, or cool, suitable for users who prefer cool colors.

[0095] In this embodiment, the key feature information of ambient light obtained by the previous cluster analysis is further sorted out, including the classification results of ambient light intensity, the clustering type of color temperature, etc., and this information is converted into a feature vector form suitable for input into a graph neural network (GNN); user visual feature data that matches the current ambient light, such as pupil size, blinking frequency, etc., are extracted from the stored data; at the same time, the user's current fatigue information is obtained through camera monitoring; the user's visual feature data is also converted into a suitable vector representation, which is compatible with the ambient light feature vector in terms of dimension and data format. The ambient light features and user visual features are respectively used as nodes in the graph, and the edges of the graph are constructed according to the correlation between them; for example, ambient light intensity and color Temperature may affect the user's pupil size and blinking frequency, so edges can be established between the corresponding nodes; assign appropriate attributes to the nodes and edges in the graph, the node attributes are feature vectors, and the edge attributes can represent the correlation strength between features; select a suitable GNN architecture (such as GCN, GAT, etc.) and use the constructed graph as input; use historical data to train GNN, and by continuously adjusting the network parameters, enable GNN to learn the complex correlation pattern between ambient light features and user visual features; after multi-layer propagation and calculation of GNN, the ambient light features and user visual features are fused to obtain a comprehensive feature representation; obtain the fused feature vector from the output layer of GNN, which contains the ambient light and user visual features. The combined information of visual features; the fused feature vector is used as the state of reinforcement learning, and the scope and dimension of the state space are clarified; the action space is defined to determine the adjustable parameters of the display, such as brightness, color temperature, contrast, etc., and the different value combinations of these parameters are defined as the action space of reinforcement learning; a deep Q network (DQN) is constructed, including an input layer, a hidden layer, and an output layer; the input layer receives state information, and the output layer outputs the Q value corresponding to each action; the network parameters of the DQN are randomly initialized, and a target network is created with the same structure and initial parameters as the DQN; in the current state, an action (i.e., a set of adjustment parameters of the display) is selected using the DQN network and the action is applied to the display; by taking a picture The camera continuously monitors the user's fatigue and uses changes in fatigue as a reward signal. If the user's fatigue decreases, a positive reward is given; if fatigue increases, a negative reward is given. The current state, action, reward, and next state are stored in the experience replay buffer. A batch of data is randomly sampled from the experience replay buffer to train the DQN network. Based on the sampled data, the loss function of the DQN network (such as mean squared error) is calculated, and the parameters of the DQN network are updated using an optimization algorithm (such as the Adam optimizer). The parameters of the target network are regularly updated to keep them consistent with the parameters of the DQN network. When encountering a new user, a small amount of ambient light and visual feature data of the user, as well as corresponding user fatigue feedback, is collected.Using a meta-learning algorithm (MAML), based on a trained DQN network and a small amount of new user data, the network parameters are quickly adjusted to adapt to the new user's visual characteristics. By iteratively updating the network parameters, the DQN network can provide personalized display adjustment parameter recommendations for the new user. The current state, a fusion of ambient light and user visual characteristics, is input into the meta-learning-adapted DQN network. Based on the Q value output by the DQN network, the action with the largest Q value is selected, and the display parameters corresponding to this action are the calculated optimal adjustment parameters.

[0096] In this embodiment, a virtual screen model is constructed in a digital space based on the physical parameters of the display screen, the current key feature information of the ambient light, and the user's visual feature data. A virtual screen model with the same properties and characteristics as the real display screen is constructed; the model can simulate the display effects under different screen parameters (brightness, color temperature, contrast, etc.); the optimal adjustment parameters are input into the virtual screen model to simulate the display effect after the display screen is adjusted; a generative adversarial network (GAN) is used to combine the user satisfaction feedback on different display effects in historical data to predict the user satisfaction under the current adjustment parameters; the GAN generator generates simulated user satisfaction data, and the discriminator judges the authenticity of the generated data. Through continuous adversarial training, the prediction accuracy is improved. The accuracy of the measurement is improved; the optimal adjustment parameters evaluated by digital twin simulation are transmitted from the local device or cloud to the control module of the display; the control module of the display adjusts the brightness, color temperature, contrast, etc. of the display in real time according to the received optimal adjustment parameters, so that the display presents the expected display effect; after the display adjustment is completed, the sensors and monitoring programs on the local device are started to prepare to collect the interaction data between the user and the display; the type of interaction data that needs to be monitored is determined, such as false touch rate, strength and frequency of touching the screen, and dwell time for reading text; the local device collects the interaction data between the user and the display in real time, and preliminarily organizes and stores this data; the collected interaction data is cleaned to remove noise and outliers ; Associate the interaction data with the current screen parameters to construct a data set for causal inference analysis; Causal analysis uses a causal inference model to analyze the causal relationship between interaction data and screen parameters; For example, through methods such as propensity score matching and instrumental variable method, determine which screen parameter adjustments lead to changes in interaction data, and the size of the causal effect of such changes; Based on the results of the causal analysis, determine whether the current screen parameter adjustment is over-adjusted; If it is found that the adjustment of certain parameters has a negative impact on the interaction data, such as a significant increase in the false touch rate, mark these parameters in time; The local device performs preliminary analysis and processing on the collected interaction data to extract key features; Based on the analysis results of the causal inference model , fine-tune the screen parameters in a small range on the local device to respond to the user's interactive feedback in real time and reduce adjustment delays; upload data to the cloud. The local device uploads the processed interaction data and key features to the cloud server; at the same time, upload relevant information of the local device during the adjustment process, such as changes in adjustment parameters, predicted results of user satisfaction, etc.; the cloud server receives data from multiple local devices, performs data aggregation and global analysis; uses these global data to optimize the model, including digital twin simulation models, causal inference models, etc., to improve the generalization ability and accuracy of the model; the cloud server sends the optimized model to each local device, and the local device updates the model to provide more accurate guidance for subsequent display adjustments.

[0097] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely preferred examples of the present invention and are not intended to limit the present invention. Various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and improvements fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A display screen adjustment system based on ambient light, characterized in that: The system includes: The data acquisition module is used to collect the initial data of the ambient light in real time using the ambient light sensor and obtain the visual feature data of the user's historical records; A classification module is used to classify the ambient light intensity based on the initial data of the ambient light using a support vector machine method; A clustering module is used to perform cluster analysis on the color temperature of the ambient light based on the initial data of the ambient light using a clustering algorithm to determine the color temperature type of the ambient light; Integration module, used to integrate the results of the classification module and clustering module to obtain the key feature information of the ambient light; The calculation module is used to retrieve the user's visual feature data that matches the current ambient light based on the key feature information of the ambient light, build a multimodal fusion model based on Transformer-GNN, optimize and improve the model's adaptive attention mechanism, calculate the comprehensive feature representation under dynamic weight distribution through the model, and calculate the optimal adjustment parameters of the display screen, and transmit the optimal adjustment parameters to the control module of the display screen; The control module is used to adjust the display screen in real time according to the optimal adjustment parameters, and continuously monitor the interaction data between the user and the display screen after the display screen adjustment is completed.

2. The display screen adjustment system based on ambient light according to claim 1, wherein: The data acquisition module includes: Build a submodule to build a federated learning network between multiple distributed devices and a central server to obtain the visual feature data of user history records collected locally by each distributed device; The setup submodule is used to set the initial sampling frequency, the initial state vector and the covariance matrix of the Kalman filter. The ambient light sensors of each distributed device collect initial ambient light data according to the initial sampling frequency. The Kalman filter algorithm is used to predict the change trend and predicted value of the ambient light parameters at the next moment through the state transition equation based on the historically collected ambient light data and the state estimation value before the current moment. The comparison submodule is used to compare the currently collected ambient light data with the predicted value, calculate the actual change in ambient light, combine the sampling time interval to obtain the rate of change of ambient light, and dynamically adjust the sampling frequency according to the rate of change of ambient light.

3. The display screen adjustment system based on ambient light according to claim 1, wherein: The classification module includes: The normalization submodule is used to extract features related to the ambient light intensity classification from the initial ambient light data, combine them into a feature vector, perform dimensionality reduction on the feature vector using principal component analysis, and normalize the feature vector after dimensionality reduction; The mapping submodule is used to map the normalized data to the high-dimensional Hilbert space through the quantum kernel function, encode it into a quantum state, measure the quantum state, and output the quantum kernel matrix; The solution submodule is used to build an SVM model based on the quantum kernel matrix, and use optimization algorithms such as gradient descent to solve the dual problem of SVM to obtain the optimal classification hyperplane; The classification submodule is used to classify the ambient light intensity based on the classification hyperplane, determine the ambient light intensity category to which it belongs, and output the classification result.

4. The display screen adjustment system based on ambient light according to claim 1, wherein: The clustering module includes: The extraction submodule is used to extract color temperature data from the initial data of ambient light and obtain the time series information of color temperature according to the collected timestamp; The initialization submodule is used to extract and cluster potential features using the DEC algorithm. The parameters of the GMM are initialized based on the results of the DEC clustering. The time series information is input into the LSTM model. The LSTM model is used to predict the color temperature data. Based on the prediction results and the current color temperature data, the GMM is used to dynamically adjust the cluster center. The determination submodule is used to integrate the cluster labels obtained by DEC clustering and the cluster center information adjusted by time-series perception clustering to determine the color temperature type of the ambient light.

5. The display screen adjustment system based on ambient light according to claim 4, characterized in that: The initialization submodule includes: A design unit is used to design an autoencoder network, which includes an input layer, a hidden layer, and an output layer, input the color temperature data into the autoencoder, and extract the output of the hidden layer as the potential feature of the color temperature data; The iterative unit is used to cluster the extracted potential features using the DEC algorithm, initialize the cluster centers, iteratively update the cluster centers and cluster assignments, minimize the clustering loss function until convergence, and thus divide the color temperature data into different categories.

6. The display screen adjustment system based on ambient light according to claim 1, wherein: The calculation module includes: The data preparation submodule is used to extract user visual feature data that matches the current ambient light and convert the key feature information of the ambient light and the user visual feature data into the feature vector form of the graph neural network; The fusion submodule is used to perform multimodal fusion network modeling on the output of the data preparation submodule to obtain the fused feature vector; The training submodule is used to use the fused feature vector as the state of reinforcement learning, determine the adjustable parameters of the display screen to obtain the action space of reinforcement learning, build the DQN network, and perform reinforcement learning training on the DQN network; The optimization submodule is used to dynamically fine-tune the parameters of the reinforcement learning model output by the training submodule, update the Q-value mapping relationship of the model in real time through online learning strategies, and perform incremental optimization of the model based on user interaction data feedback; The selection submodule is used to input the fused feature vector into the trained DQN network and select the action with the largest Q value according to the Q value output by the DQN network. The display screen parameters corresponding to this action are the calculated optimal adjustment parameters.

7. The display screen adjustment system based on ambient light according to claim 6, characterized in that: The fusion submodule includes: A construction unit, configured to use the key feature information of ambient light and the user's visual feature data as nodes in a graph, and to construct edges of the graph based on the association relationship; The adjustment unit is used to use the constructed graph as the input of the graph neural network, use historical data to train the graph neural network, and continuously adjust the network parameters to enable the graph neural network to learn the complex correlation pattern between ambient light characteristics and user visual characteristics; The first computing unit is used to fuse the key feature information of ambient light and the user's visual feature data through multi-layer propagation and calculation of the graph neural network to obtain a comprehensive feature representation.

8. The display screen adjustment system based on ambient light according to claim 6, characterized in that: The DQN network includes an input layer, a hidden layer, and an output layer, wherein the output layer outputs the Q value corresponding to each action.

9. The display screen adjustment system based on ambient light according to claim 6, wherein: The training submodule includes: The selection unit is used to select an action using the DQN network in the current state and apply the action to the display screen; A continuous monitoring unit is used to continuously monitor the user's fatigue, use the change in the user's fatigue as a reward signal, and store the current state, action, reward, and next state in the experience replay buffer; The second computing unit is used to randomly sample a batch of data from the experience replay buffer for training the DQN network, calculate the loss function of the DQN network based on the sampled data, and use the Adam optimizer to update the parameters of the DQN network to complete the reinforcement learning training of the DQN network.

10. The display screen adjustment system based on ambient light according to claim 1, wherein: The control module includes: A collection submodule is configured to collect, after the display screen is adjusted, user interaction data between the display screen and the user through sensors on the local device, where the interaction data includes at least false touch rate, touch force and frequency, and dwell time on reading text; The storage submodule is used to collect the interaction data between the user and the display screen in real time, and to organize and store the interaction data.

Images