Dynamic backlight compensation optimization method for multi-user screen interaction

Abstract

This invention provides a dynamic backlight compensation optimization method for multi-user screen interaction, relating to the field of screen backlight control technology. This invention introduces real-time acquisition and analysis of multi-user behavior data and gaze data, combines a long short-term memory network model for time series prediction, and achieves comprehensive weight evaluation of each marked area of ​​the screen through collaborative weight calculation and weighted fusion. This not only accurately identifies the interaction needs and gaze distribution of multiple users and adjusts backlight brightness in real time, achieving the dual goals of visual optimization and energy consumption optimization, but also improves the intelligence level of the screen display system and the user experience.

Description

Technical Field

[0001] This invention relates to the field of screen backlight control technology, specifically to a dynamic backlight compensation optimization method for multi-user screen interaction. Background Technology

[0002] In multi-user interactive environments, dynamically adjusting screen backlight brightness based on the real-time interactive behavior and gaze distribution of different users has become an important research direction for improving user experience and optimizing energy efficiency. Existing backlight control technologies mainly rely on static measurements of a single user or simple preset modes, which often fails to accurately respond to the personalized needs of each user in multi-user scenarios. In addition, traditional backlight adjustment methods lack the ability to analyze user dynamic behavior and gaze changes in real time, resulting in lag and inaccuracy in backlight adjustment, making it impossible to achieve fine-grained visual optimization and comprehensive energy consumption control.

[0003] In the prior art, publication number CN117690385A, entitled "A Backlight Optimization Method, a Backlight Optimization Device, an Electronic Device, and a Readable Storage Medium," the backlight optimization method includes: obtaining the current color gamut mode and current backlight value of the screen when an application is adapted to dark mode, the screen displaying the interface of the application; determining the backlight region type where the current backlight value is located, the backlight region type including low backlight region and high backlight region; and optimizing the backlight of the screen based on the backlight region type where the current backlight value is located, the current color gamut mode, and the current backlight value.

[0004] Existing technologies have significant shortcomings when dealing with multi-user screen interactions. First, most existing backlight optimization methods only consider the interaction behavior of a single user, ignoring the complexity of screen display requirements under simultaneous multi-user operation, leading to uneven resource allocation and inconsistent user experience. Second, traditional methods often use simple weighting or ignore the dynamic changes in gaze trajectory when processing user gaze data, failing to fully capture the user's real-time focus and interaction intentions. Furthermore, existing technologies lack intelligent strategies for energy consumption optimization, failing to dynamically adjust backlight brightness based on changes in screen content and user behavior, thus affecting the device's energy efficiency and battery life.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a dynamic backlight compensation optimization method for multi-user screen interaction, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A dynamic backlight compensation optimization method for multi-user screen interaction includes the following steps:

[0009] Step S1: Collect in real time the screen interaction behavior data of the operating user on each marked area of ​​the target screen during the current monitoring time period, as well as the gaze data of multiple viewing users; preprocess the collected data to generate user behavior data matrix and user gaze data matrix respectively;

[0010] Step S2: Input the user behavior data matrix of the operating user into the Long Short-Term Memory network model, analyze the time series characteristics, and predict the behavior weight matrix of each marked area on the target screen at the next time step;

[0011] Step S3: Obtain the user gaze data matrix from multiple viewing users and perform collaborative weight calculation to generate a collaborative gaze weight distribution matrix;

[0012] The behavior weight matrix and the collaborative gaze weight distribution matrix are weighted and fused to generate a fused weight model, and the comprehensive weight level is marked for each marked area of ​​the target screen based on the fused weight model.

[0013] Step S4: Analyze the comprehensive weight level marking results of each marked area on the target screen and generate an initial backlight brightness adjustment matrix;

[0014] Step S5: Calculate the energy consumption of each marked area of ​​the target screen during the current monitoring period, and calculate the total energy consumption of the target screen. Compare the total energy consumption with the preset adjustment threshold, and compensate and optimize the initial backlight brightness adjustment matrix based on the comparison results to generate the final backlight brightness adjustment matrix.

[0015] Compared with the prior art, the beneficial effects of the present invention are:

[0016] By introducing real-time collection and analysis of multi-user behavior data and gaze data, combined with a Long Short-Term Memory (LSTM) network model for time series prediction, and through collaborative weight calculation and weighted fusion, a comprehensive weight evaluation of each marked area on the screen is achieved. This method can not only accurately identify the interaction needs and gaze distribution of multiple users, but also adjust the backlight brightness in real time, achieving the dual goals of visual optimization and energy consumption optimization, thus improving the intelligence level of the screen display system and the user experience. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0019] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0020] Example 1:

[0021] Please see Figure 1 The present invention provides a technical solution:

[0022] A dynamic backlight compensation optimization method for multi-user screen interaction includes the following steps:

[0023] Step S1: Collect in real time the screen interaction behavior data of the operating user on each marked area of ​​the target screen during the current monitoring time period, as well as the gaze data of multiple viewing users; preprocess the collected data to generate user behavior data matrix and user gaze data matrix respectively;

[0024] Further explanation: The target screen is divided into a set of marked regions, denoted as {1, 2, ..., i, ..., n}, where i represents the index of the marked region on the target screen and n is the total number of marked regions; each marked region corresponds to an independent data acquisition unit;

[0025] Screen interaction behavior data includes screen interaction pressure sensing data and finger contact area change data;

[0026] The following information about the user is collected in real time through hardware devices that can sense user behavior (touchscreen pressure sensor and area sensor):

[0027] Screen interaction pressure sensing data: Records the pressure applied by the user on each marked area of ​​the target screen; the pressure sensor can directly collect and generate pressure data sequences;

[0028] Finger contact area change data: Records changes in the finger contact area when the user interacts with the screen (such as area differences caused by finger tilt or different force); collected in real time through the area sensing sensor of the touch screen hardware;

[0029] Denoising and trajectory correction based on Kalman filtering algorithm are performed on the collected user behavior data.

[0030] Noise reduction: The Kalman filter algorithm can capture the trajectory of the user's finger swipe path and eliminate noise caused by sampling errors or external interference.

[0031] Trajectory correction: The actual user interaction trajectory is calibrated through filtering algorithms to ensure the accuracy and reliability of the data.

[0032] The denoised screen interaction data is organized in matrix form, defined as user behavior data matrix U = [P] i A i ];

[0033] Among them, P i It is the screen interaction pressure sensing data of the i-th marked area; A i This represents the change in the finger contact area of ​​the i-th marked region;

[0034] The following parameters are set to be collected multiple times within the current monitoring time period, and the data collected each time is collected within the preset time period; and each parameter is processed to be dimensionless, and the output value is within the same range of (0, 1).

[0035] Calculate screen interaction pressure sensing data P i The calculation formula is as follows:

[0036]

[0037] Among them, P i This is the screen interaction pressure sensing data for the i-th marked area, and the output value range is limited to (0, 1) through normalization; FN i It is the magnitude of the force applied by the user in the i-th marked area, measured by a pressure sensor; W p This is the pressure-sensitive weight of the screen material, obtained through screen calibration experiments, and is measured in dimensionless numbers; AS i It is the size of the finger contact area, measured by an area sensor; μ1 is to ensure that the output value range is limited to (0, 1);

[0038] When P i The closer it is to 1, the greater the effective pressure applied by the user in the i-th marked area, reflecting a strong need for the screen area to brighten.

[0039] When P i The closer it gets to 0, the smaller the effective pressure applied by the user in the i-th marked area, and the i-th marked area remains dark.

[0040] Increase FN i Or reduce AS i This will lead to P i An increase indicates that users need clearer interactive feedback;

[0041] Calculate the change in finger contact area data A i The calculation formula is as follows:

[0042]

[0043] Among them, A i This represents the change in finger contact area in the i-th marked region, and the output value range is limited to (0, 1) through normalization; Nc i VA is the number of contacts in the i-th marked area, counted by a multi-contact sensor; t,i It is to operate the user's finger on Nc i Average contact area change for each number of contacts; T1 i is the dwell time of the finger in the i-th marked region; μ2 is to ensure that the output value range is limited to (0, 1);

[0044] When A i The closer the value is to 1, the larger the contact area of ​​the user in the i-th marked area;

[0045] When A i The closer it is to 0, the smaller the contact area of ​​the user in the i-th marked area;

[0046] Increase Nc i or VA t,i It will make A i An increase indicates active user interaction.

[0047] Furthermore, the gaze data includes the duration of each viewing user's gaze at each marked area on the target screen, gaze frequency, pupil diameter variation data, blink frequency, and user gaze trajectory index;

[0048] During the current monitoring period, the following gaze data of multiple users on the target screen are collected in real time using high-precision eye-tracking devices or camera systems:

[0049] Gaze duration: Records the duration of each user's gaze on each marked area of ​​the target screen, obtained through the device time statistics module;

[0050] Fixation frequency: Calculates the fixation frequency of the viewing user on each marked area of ​​the target screen, obtained through an eye-tracking data analysis engine;

[0051] Pupil diameter variation data: Records the dilation or constriction of the pupil due to changes in emotion and interest, acquired through infrared tracking equipment;

[0052] Blink frequency: Collects the time interval and frequency of the user's blinks, which are obtained in real time through the camera data analysis module;

[0053] User gaze trajectory index: This index records the rotation direction and angle of the user's eyes, as well as the sequence of areas the user looks at on the screen in turn, and is generated by an eye-tracking device.

[0054] Cluster analysis was performed on the gaze data collected from multiple viewing users; in this embodiment, the cluster analysis was based on the K-means algorithm:

[0055] Clustering objective: Identify the area of ​​focus of the user's gaze on the target screen, and quantify the degree of attention to each area by processing information such as gaze duration;

[0056] Generate a heatmap of line-of-sight distribution; cluster analysis results are output in the form of a heatmap, with color mappings indicating the level of attention to different areas.

[0057] Based on the gaze distribution heatmap results, a user gaze data matrix V = [T] is generated. i ,F i D i ,R i ,S i ];

[0058] Among them, T i F is the fixation duration for the i-th marked region; i D is the fixation frequency of the i-th marked region; i R is the change in pupil diameter in the i-th marked region; i It is the user's gaze trajectory index; S i It is the blink frequency of the i-th marked region;

[0059] Define gaze duration T i The formula is as follows:

[0060]

[0061] Among them, T i T is the fixation duration for the i-th labeled region, and its output range is limited to (0, 1) through normalization. pi T is the absolute time spent continuously observing the i-th marked region during the entire monitoring period; maxiIt is the maximum gaze time of the user on the screen display of the i-th marked area; μ3 is a constant to prevent the denominator from being zero;

[0062] When T i The closer a value is to 1, the longer the user's gaze duration.

[0063] Calculate pupil diameter variation data D i The formula is:

[0064]

[0065] Among them, D i This represents the pupil diameter change in the i-th labeled region, and the output value range is limited to (0, 1) through standardization; P max It is the pupil diameter when the attraction of the i-th marked region is the highest; P min It is the pupil diameter when the user loses interest in a region; P neutral This is the standard value for the user's pupil diameter; μ4 is a correction term used to ensure that the value range is (0, 1);

[0066] D i The closer the value is to 1, the greater the impact of the i-th marked region on the change in the pupil diameter of the viewing user;

[0067] Calculate the gaze frequency F of the i-th marked region. i The formula is as follows:

[0068]

[0069] ZC i It represents the number of times a user gazes at the i-th marked region during the current monitoring time period; N total It is the total number of times a user gazes at all marked areas during the current monitoring period; F i The output range is limited to (0, 1) through normalization; when F i The closer it is to 1, the more frequently the viewing user stares at the i-th marked area, and the more attractive the content of the i-th marked area is;

[0070] Calculate the user's gaze trajectory index R i The calculation formula is as follows:

[0071]

[0072] Among them, R i It is the index of the user's gaze trajectory on the i-th marked area, and the output value range is limited to (0, 1) through standardization.

[0073] △θ iIt is the total change in the horizontal rotation angle of the user's eyeball in the i-th marked area; △φ i It is the total change in the vertical rotation angle of the user's eyeball in the i-th marked area; It is the number of adjacent marked regions where the gaze order changes; that is, the total number of gaze order points between marked regions for the user's gaze; W r It is a complexity weighting factor for gaze switching; it is used to express the relationship between the complexity of the viewer's focus direction and the attractiveness of the screen content, and is an empirical calibration value.

[0074] μ5 is a constant to prevent the denominator from being zero; in this embodiment, it is taken as 0.01.

[0075] Analyze the order in which the user's gaze switches between marked areas, and count the number of adjacent marked area points in the i-th marked area. Determine the sequence point from which the viewing user switches from the i-th marked region to the adjacent marked region. The sequence point represents the order in which the viewing user switches to the adjacent marked region for gaze.

[0076] When R i The closer a value is to 1, the more frequently the viewer's gaze switches between the i-th marked area;

[0077] When △θ i An increasing value indicates that the viewer's gaze shifts more horizontally across the screen, suggesting that the content's appeal is primarily characterized by a horizontal layout. In this case, R... i Increase synchronously;

[0078] When △φ i An increase in R indicates that the viewer has made a significant vertical eye rotation on the screen, suggesting that the content's appeal is primarily based on vertical information. i Increase synchronously;

[0079] When W r Increase or Decrease: This indicates increased weighting and more pronounced regional complexity; decreased sequence points lead to longer user dwell time on individual regions; the combined effect of both increases R. i Increase synchronously.

[0080] Output the user behavior data matrix U and the user gaze data matrix V to a unified data interface;

[0081] User behavior data matrix U = [P] i A i This indicates the real-time behavior of the user interacting with the screen.

[0082] User gaze data matrix V = [T i ,F i D i ,Ri ,S i [] indicates the dynamic visual behavior of the user interacting with the marked area on the screen.

[0083] Step S2: Input the user behavior data matrix of the operating user into the Long Short-Term Memory network model, analyze the time series characteristics, and predict the behavior weight matrix of each marked area on the target screen at the next time step;

[0084] Further explanation: The behavior weight matrix learns user behavior characteristics and assigns weight values ​​to each marked region, which serve as a reference for the optimization priority of each marked region on the target screen;

[0085] The real-time collected user behavior data matrix is ​​sorted according to timestamps to form a time series dataset; the data points at each time step include P of all n labeled regions. i and A i data;

[0086] P i and A i The values ​​are all within the same range of (0, 1) to improve the training efficiency and prediction accuracy of the LSTM model;

[0087] Set the time window length to T time steps, and divide the time series data into multiple input samples. Each sample contains continuous time step data of length T, which are used as input to the LSTM model.

[0088] Sample k =[X k ,X k+1 ,…,X k+T-1 ];

[0089] Among them, Sample k is the training sample, an input sequence consisting of data from consecutive time steps, used as input to the LSTM model; T is the time step of the time window length, which is selected by cross-validation to determine the best-performing time window on the validation set; k is the sample index, representing the starting position of the currently generated training sample in the time series; k∈{0,1,2,…,MT-1}; M is the total number of time steps in the time series dataset;

[0090] The LSTM model adopts a multi-layer bidirectional Bi-LSTM model architecture to fully capture the dependencies between time series data and improve prediction accuracy.

[0091] The specific model structure is as follows:

[0092] Input layer: Accepts input data of (T, 2n), where 2n represents P for each labeled region. i and A idata;

[0093] Bidirectional LSTM layer: Two bidirectional LSTM layers are set up, each containing 128 neurons, to enhance the model's ability to learn complex time dependencies.

[0094] Fully connected layer: The output of the bidirectional LSTM is mapped to the output space of the behavior weight matrix through the fully connected layer.

[0095] Activation function: The Sigmoid activation function is used to ensure that the output weight values ​​are confined to the range (0, 1). The Sigmoid activation function is expressed as follows:

[0096]

[0097] Loss function: Mean squared error (MSE) is chosen to measure the difference between the predicted weight matrix and the true weight matrix;

[0098] Optimizer: The Adam optimizer is used because it performs well when dealing with large-scale data and non-convex optimization problems.

[0099] Learning rate: Set the initial learning rate to 0.001 and use a learning rate decay strategy to improve training stability.

[0100] Batch size: Select 64 to ensure training efficiency while stabilizing model convergence.

[0101] Training rounds: Set to 100 rounds and monitor overfitting on the validation set.

[0102] The preprocessed time series dataset was divided into training, validation and test sets in an 80:10:10 ratio to ensure that the model has good generalization ability.

[0103] The training process is as follows:

[0104] Initialize model parameters: Randomly initialize network weights, using He initialization to facilitate gradient flow;

[0105] Forward propagation: Input training samples and generate the predicted behavior weight matrix through bidirectional LSTM layers and fully connected layers;

[0106] Loss calculation: Use MSE to calculate the error between the predicted value and the actual value;

[0107] Backpropagation and optimization: The gradient is calculated using the backpropagation algorithm, and the model weights are updated using the Adam optimizer;

[0108] Validation monitoring: After each training cycle, evaluate the loss on the validation set and apply an early stopping mechanism to prevent overfitting;

[0109] Model saving: Record the model parameters that perform best on the validation set to ensure that the final model has the best performance;

[0110] The latest user behavior data matrix at time step T [X] k ,X k+1 ,…,X k+T-1 Input the trained Bi-LSTM model;

[0111] The predicted behavior weight matrix W is generated through model calculation. pred =[w1,w2,…,w n ], where w i ∈(0,1) represents the predicted weight value of the i-th labeled region;

[0112] w i This indicates the importance of the interaction of the i-th marked region in the next time step; a higher value indicates a higher priority.

[0113] Behavioral weight matrix W pred =[w1,w2,…,w n This will serve as a priority reference for subsequent dynamic backlight optimization of the screen, guiding the brightness adjustment strategy for each area;

[0114] The predicted behavior weight matrix W pred =[w1,w2,…,w n Output in structured data format, containing the weight values ​​w for each labeled region. i The structured data formats in this embodiment include JSON and CSV.

[0115] The historical predicted behavior weight matrix and actual user behavior data are stored in an SQL or NoSQL database for subsequent model optimization and behavior analysis.

[0116] This embodiment has made significant improvements and innovations in the following aspects:

[0117] Multi-feature fusion: By comprehensively utilizing screen interaction pressure sensing data and finger contact area change data, the LSTM model learns multi-dimensional and interaction time characteristics to provide more comprehensive behavior weight prediction.

[0118] Behavioral weight matrix W pred =[w1,w2,…,w n The prediction result is obtained by nonlinearly mapping multidimensional time series data using an LSTM model, representing the interaction priority of each labeled region in the next time step.

[0119] Physical significance: It reflects the current trend and future needs of users' interaction with the screen, guides the real-time adjustment of the dynamic backlight optimization system, and achieves intelligent response.

[0120] Step S3: Obtain the user gaze data matrix from multiple viewing users and perform collaborative weight calculation to generate a collaborative gaze weight distribution matrix;

[0121] The behavior weight matrix and the collaborative gaze weight distribution matrix are weighted and fused to generate a fused weight model, and the comprehensive weight level is marked for each marked area of ​​the target screen based on the fused weight model.

[0122] Further explanation: During the predetermined monitoring period, gaze data of each user towards each marked area on the target screen will be continuously collected;

[0123] The gaze data from all users is aggregated and integrated to construct a gaze data matrix V = [T] i ,F i D i ,R i ,S i ]; Set each row of the gaze data matrix to correspond to a marked area on the target screen, and the columns to correspond to different gaze data indicators;

[0124] Calculate the collaborative gaze weight distribution matrix W gaze =[w g1 ,w g2 ,…,w gn ], where w gj ∈(0,1) represents the collaborative gaze weight value of the j-th marked region; j∈{1,2,…,n};

[0125] W gaze =[w g1 ,w g2 ,…,w gn The steps to obtain ] are as follows:

[0126] 1.1) Feature weight assignment: Based on the technical importance of the gaze data, different weight coefficients (α) are assigned to each indicator. T ,α F ,α D ,α R ,α S ), satisfying α T +α F +α D +α R +α S =1;

[0127] The method for selecting weighting coefficients is to determine the optimal weight values ​​by using a weighted average method, combining expert knowledge and experimental data. This embodiment provides...

[0128] Determine the line-of-sight data matrix V = [Ti ,F i D i ,R i ,S i All the index values ​​in the table are within the range of (0, 1);

[0129] 1.2) Collaborative weight calculation: For each marked region j, calculate the collaborative gaze weight value w. gj As a weighted sum of all indicators:

[0130] w gj =α T ·T j +α F ·F j +α D ·D j +α R ·R j +α S ·S j ;

[0131] Where T j F j D j R j S j These are the fixation duration, fixation frequency, pupil diameter change, user gaze trajectory index, and blink frequency for the j-th marked region, respectively.

[0132] 1.4) Weight Standardization: The collaborative gaze weight distribution matrix W... gaze =[w g1 ,w g2 ,…,w gn Standardization is performed using the Softmax function to ensure that the sum of the weight values ​​is 1;

[0133] The behavior weight matrix and the collaborative gaze weight distribution matrix are weighted and fused to obtain the fused weight model W. fusion =[w f1 ,w f2 ,…,w fn ];

[0134] W fusion =[w f1 ,w f2 ,…,w fn The steps to obtain ] are as follows:

[0135] w fj =β·w j +(1-β)·w gj ;

[0136] Where β∈(0,1) is the fusion coefficient; determined based on system requirements and experimental results, in this embodiment β=0.6, indicating that the behavioral weights account for a larger proportion in the fusion; w fj It is the fusion weight value of the labeled region j;

[0137] The fusion weight W fusion Normalization ensures that the weight values ​​are within the range of (0, 1);

[0138] Based on the fusion weight value w fj The output value is used to classify the overall weight level using a fixed threshold method, setting the overall weight level as high priority, medium priority, and low priority. fj The larger the value, the higher the priority of the marked region j; this embodiment uses a fixed threshold method for division:

[0139] High priority: w fj ≥0.66;

[0140] Medium priority: 0.33≤w fj <0.66;

[0141] Low priority: w fj <0.33;

[0142] The aforementioned fixed threshold method is specifically determined using the fuzzy hierarchical analysis method (FAHP);

[0143] When w fj The closer the value is to 1, the higher the interaction and attention of the marked area j, and the more attractive the marked area j is to users;

[0144] The dynamic backlight optimization system will prioritize enhancing the brightness or other visual attributes of the area to improve user experience and interaction efficiency;

[0145] When w fj The closer a value is to 0, the lower the interaction and attention level of the marked region j, and the less attractive the marked region j is to users.

[0146] Step S4: Analyze the comprehensive weight level marking results of each marked area on the target screen and generate an initial backlight brightness adjustment matrix;

[0147] Further explanation: The initial backlight brightness adjustment matrix is ​​used to initially control the luminous intensity of the LEDs in each marked area on the target screen;

[0148] Brightness changes are achieved by adjusting the electrical signal strength of the LEDs through the screen control unit;

[0149] Let high priority, medium priority, and low priority be denoted as G1, G2, and G3, respectively;

[0150] Set the adjustable range of backlight brightness for each marked area on the target screen to [Ld1, Ld2], in %, and Ld1 = Ld2 × 20%;

[0151] In this embodiment: the adjustable range of the backlight brightness of the target screen [Ld1, Ld2] is set to [20%, 100%]; the adjustable range of the backlight brightness is adaptively adjusted according to the model of the target screen, which will not be elaborated here;

[0152] The backlight brightness of high-priority weights is set to the maximum value of the adjustable backlight brightness range;

[0153] The backlight brightness of the medium priority weight is set to the medium value of the adjustable backlight brightness range;

[0154] The backlight brightness of low-priority weights is set to the lowest value in the adjustable range of backlight brightness;

[0155] In this embodiment, the maximum, median, and minimum values ​​are the maximum, median, and minimum values ​​of the adjustable backlight brightness range, respectively.

[0156] Assign a corresponding relative brightness ratio coefficient L based on the priority of each marked region j. j :

[0157]

[0158] Among them, dJ j It is an abbreviation for the priority level of the marked region j;

[0159] if dJ j =G1 indicates that when the marked region j is of high priority, the relative brightness scaling factor L will be... j Characterized as 1.0; L j =1 indicates that the initial backlight brightness of the marked area j is set to the maximum brightness of the adjustable backlight brightness range [Ld1, Ld2], which is Ld2;

[0160] if dJ j =G2 indicates that when the marked area j is of medium priority, the relative brightness scaling factor L will be... j Characterized as 0.5; L j =0.5 indicates that the initial backlight brightness of the marked area j is set to half of the maximum brightness of the adjustable backlight brightness range [Ld1, Ld2].

[0161] if dJ j =G3 indicates that when the marked area j is of low priority, the relative brightness scaling factor L will be... j Characterized as 0.2; L j=0.2 indicates that the initial backlight brightness of the marked area j is set to the lowest adjustable brightness in the adjustable backlight brightness range [Ld1, Ld2], which is Ld1;

[0162] Setting L j The value range is [0, 1]; the relative brightness ratio coefficient L j Used to represent the ratio of the initial backlight brightness to Ld2;

[0163] In this embodiment: the maximum brightness of the target screen backlight is 500 cd / m²;

[0164] The actual backlight brightness corresponding to a ratio of 1 is 500 cd / m². 2 ;

[0165] The actual backlight brightness corresponding to a ratio of 0.5 is 500 × 0.5 = 250 cd / m². 2 ;

[0166] The actual backlight brightness corresponding to a ratio of 0.2 is 500 × 0.2 = 100 cd / m².

[0167] Let the initial backlight brightness adjustment matrix be denoted as M. brightness =[L1,L2,…,L n ].

[0168] The steps to adjust the luminous intensity are as follows:

[0169] The brightness adjustment command is sent to the screen control unit via a wired or wireless communication interface.

[0170] High-priority areas: control the luminous intensity of the corresponding LEDs to the maximum value to improve brightness;

[0171] Medium priority area: Control the light intensity of the corresponding LED to a medium value, maintaining medium brightness;

[0172] Low priority area: Control the light intensity of the corresponding LED to the lowest value, and maintain low brightness.

[0173] Step S5: Calculate the energy consumption of each marked area of ​​the target screen during the current monitoring period, and calculate the total energy consumption of the target screen. Compare the total energy consumption with the preset adjustment threshold, and compensate and optimize the initial backlight brightness adjustment matrix based on the comparison results to generate the final backlight brightness adjustment matrix.

[0174] Furthermore, the backlight brightness adjustment matrix M is adjusted based on the initial backlight brightness. brightness =[L1,L2,…,L n ] Calculate the energy consumption of each marked area;

[0175] E j =JNhj ×L j ;

[0176] Among them, JNh j E is the reference energy consumption of the LED in marked area j under a unit relative brightness ratio factor; the reference energy consumption is the energy consumption of the LED at its maximum brightness based on the unit relative brightness ratio factor; j This is the actual energy consumption of the light-emitting diode in the marked area j;

[0177] Calculate the total energy consumption of the target screen. The preset adjustment threshold is set to the adjustment threshold range [E] th1 E th2 ]; E th1 With E th2 Total energy consumption E total Adjust the lower and upper limits of the threshold;

[0178] [E th1 E th2 The corresponding weights are determined using the entropy weight method;

[0179] If E th1 ≤E total ≤E th2 Keep the initial backlight brightness value unchanged;

[0180] If E total >E th2 Then reduce the initial backlight brightness value of the medium-priority and high-priority areas to make E total Back to E th1 ≤E total ≤E th2 ;

[0181] The specific operating steps are as follows:

[0182] Calculate the adjustment ratio δ down :

[0183] Let δ down δ represents the ratio of brightness reduction in medium-priority and high-priority areas. down The value range is [0, 1];

[0184] The total brightness reduction percentage required to calculate the initial backlight brightness value is:

[0185]

[0186] n midHigh This refers to the number of medium-priority and high-priority regions;

[0187] If δ down>0.4, the brightness reduction is set within the adjustable backlight brightness range [Ld1, Ld2]. 40% to prevent excessive reduction from affecting the visual effect;

[0188] Brightness adjustment is performed on medium-priority and high-priority areas using the following formula:

[0189] L′ j =L j ·(1-δ down );

[0190] Among them, L′ j It is the relative brightness ratio coefficient of the marked area j after adjustment, and ensures that the adjusted backlight brightness is within the adjustable range of backlight brightness [Ld1, Ld2].

[0191] Recalculate E total ,get

[0192] If E th1 ≤E total ≤E th2 Stop adjusting; otherwise, repeat the above steps.

[0193] If E total <E th1 This increases the initial backlight brightness value of the medium-priority and high-priority areas to make E total Increase to E th1 ≤E total ≤E th2 Location range.

[0194] The specific operating steps are as follows:

[0195] Calculate the lifting ratio δ up ;

[0196] Let δ up δ represents the brightness increase ratio between medium-priority and high-priority regions. up The value range is within the interval [0, 1].

[0197] The percentage increase in total brightness required to calculate the initial backlight brightness value is:

[0198]

[0199] If δ up >0.3, the brightness increase is set within the adjustable backlight brightness range [Ld1, Ld2]. 30% to prevent excessive increases from affecting energy consumption control;

[0200] Brightness enhancement is applied to medium-priority and high-priority regions using the following formula:

[0201] L′ j =min[L j ·(1+δ up ),1.0];

[0202] Where min[L j ·(1+δ up ),1.0] indicates from L j ·(1+δ up Choose the minimum value between 1.0 and 1.0;

[0203] And ensure that the adjusted backlight brightness is within the adjustable range [Ld1, Ld2];

[0204] Recalculate E total ,get

[0205] If E th1 ≤E total ≤E th2 Stop adjusting if necessary; otherwise, repeat the above steps.

[0206] The final backlight brightness adjustment matrix is ​​represented as M′. brightness =[L′1,L′2,…,L′ n ];

[0207] The final backlight brightness adjustment matrix is ​​represented as M′. brightness =[L′1,L′2,…,L′ n The signal strength is converted into an electrical signal strength format that the screen control unit can recognize; in this embodiment, the electrical signal strength format is a PWM signal or a DAC drive voltage.

[0208] In the above enhancement and reduction strategies, ensure that the adjusted relative brightness ratio coefficient L′ is maintained. j All values ​​must be within the range [0, 1] and must not exceed this range; if the adjusted value exceeds this range, it will be determined according to 0 or 1 in the interval [0, 1].

[0209] The Screen Control Unit (SCU) is the core hardware module or electronic system used to control the backlight brightness of a display screen. It is responsible for receiving the backlight brightness adjustment matrix, executing the brightness adjustment algorithm, and finally driving the light-emitting diodes (LEDs) or other backlight sources through electrical signals to achieve dynamic brightness changes in each marked area of ​​the target screen.

[0210] Furthermore, the input terminal of the screen control unit receives the final backlight brightness adjustment matrix, which is transmitted to the control unit via an electrical signal strength format. After parsing and processing, the control unit generates corresponding LED luminous intensity adjustment commands based on the received electrical signal strength, thereby dynamically adjusting the LED luminous intensity of each marked area on the target screen. This achieves personalized brightness adjustment and visual optimization for different areas.

[0211] This embodiment has the following beneficial effects:

[0212] The combination of comprehensive weighting levels and brightness grading: By dividing backlight brightness into high, medium and low priorities and corresponding to different brightness values ​​through comprehensive weighting levels, more detailed and flexible backlight brightness control is achieved, improving the accuracy of visual optimization.

[0213] Dynamic balance between energy consumption and visual optimization: By introducing energy consumption calculation and dynamic adjustment strategies, energy consumption can be effectively controlled while ensuring visual effects, thus achieving dual optimization of visual appeal and energy efficiency.

[0214] Multi-level brightness control matrix: By defining fixed brightness levels, a simple and efficient brightness control logic is achieved, reducing system complexity while ensuring optimization results.

[0215] It should be noted that all calculation formulas in this application employ, but are not limited to, regression analysis from machine learning algorithms to deeply analyze the collected relevant parameters and identify their natural trends and interrelationships. Specialized software, such as Python's Scikit-learn library or the R language, is used to automatically generate mathematical models that match the data. Then, cross-validation and other methods are used to objectively evaluate the model performance, and continuous feedback and optimization are combined to ensure that the created formulas truly reflect the inherent laws of the data, thereby guaranteeing their validity and accuracy, and ensuring that the calculation process conforms to the constraints of natural laws, rather than being based on artificially set rules.

[0216] The technical solution of this invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of this invention.

[0217] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0218] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0219] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A dynamic backlight compensation optimization method for multi-user screen interaction, characterized in that, The specific steps include: Step S1: Collect in real time the screen interaction behavior data of the operating user on each marked area of ​​the target screen during the current monitoring time period, as well as the gaze data of multiple viewing users; preprocess the collected data to generate user behavior data matrix and user gaze data matrix respectively; Step S2: Input the user behavior data matrix of the operating user into the Long Short-Term Memory network model, analyze the time series characteristics, and predict the behavior weight matrix of each marked area on the target screen at the next time step; Step S3: Obtain the user gaze data matrix from multiple viewing users and perform collaborative weight calculation to generate a collaborative gaze weight distribution matrix; The behavior weight matrix and the collaborative gaze weight distribution matrix are weighted and fused to generate a fused weight model, and the comprehensive weight level is marked for each marked area of ​​the target screen based on the fused weight model. Calculate the collaborative line-of-sight weight distribution matrix ,in This represents the collaborative gaze weight value for the j-th labeled region; j∈{1,2,…,n}; and the predicted behavior weight matrix is ​​obtained. , This represents the predicted weight value for the j-th labeled region; The behavior weight matrix and the collaborative gaze weight distribution matrix are weighted and fused to obtain the fused weight model. ; The steps to obtain it are as follows: ； in, The fusion coefficient; It is the fusion weight value of the labeled region j; Based on the fusion weight value The output value is used to divide the comprehensive weight level using a fixed threshold method, and the comprehensive weight levels are set as high priority, medium priority, and low priority. when The closer the value is to 1, the higher the interaction and attention of the marked area j, and the more attractive the marked area j is to users; when The closer a value is to 0, the lower the interaction and attention given to the marked region j, and the less attractive the marked region j is to users. Step S4: Analyze the comprehensive weight level marking results of each marked area on the target screen and generate an initial backlight brightness adjustment matrix; Step S5: Calculate the energy consumption of each marked area of ​​the target screen during the current monitoring period, and calculate the total energy consumption of the target screen. Compare the total energy consumption with the preset adjustment threshold, and compensate and optimize the initial backlight brightness adjustment matrix based on the comparison results to generate the final backlight brightness adjustment matrix.

2. The dynamic backlight compensation optimization method for multi-user screen interaction according to claim 1, characterized in that: The target screen is divided into a set of marked regions, denoted as {1,2,…,i,…,n}, where i represents the index of the marked region on the target screen and n is the total number of marked regions; Screen interaction behavior data includes screen interaction pressure sensing data and finger contact area change data; Denoising and trajectory correction based on Kalman filtering algorithm are performed on the collected user behavior data. The denoised screen interaction data is organized in matrix form and defined as the user behavior data matrix. ; in, It is the screen interaction pressure sensing data of the i-th marked area; It represents the change in finger contact area in the i-th marked region.

3. The dynamic backlight compensation optimization method for multi-user screen interaction according to claim 2, characterized in that: The gaze data includes the duration of each user's gaze at each marked area on the target screen, the frequency of gaze, the change in pupil diameter, the blink frequency, and the user's gaze trajectory index; Cluster analysis was performed on the gaze data collected from multiple viewing users; Generate a heatmap of line-of-sight distribution; Generate a user gaze data matrix based on gaze distribution heatmap results. ; in, It is the gaze duration of the i-th marked region; It is the gaze frequency of the i-th marked region; It represents the change in pupil diameter in the i-th marked region; It is the user's gaze trajectory index; It is the blink frequency of the i-th marked region; Calculate the user's gaze trajectory index The calculation formula is as follows: ； in, It is the index of the user's gaze trajectory on the i-th marked area, and the output value range is limited to (0,1) through standardization. It is the total change in the horizontal rotation angle of the user's eyeball in the i-th marked area; It is the total change in the vertical rotation angle of the user's eyeball in the i-th marked area; It is the number of adjacent marked regions that switch gaze order; It is a constant to prevent the denominator from being zero. It is a weighting factor for the complexity of gaze switching; when The closer the value is to 1, the more frequently the user switches their gaze in the i-th marked area.

4. The dynamic backlight compensation optimization method for multi-user screen interaction according to claim 3, characterized in that: The behavior weight matrix learns user behavior characteristics and assigns weight values ​​to each marked region, which serve as a reference for the optimization priority of each marked region on the target screen. The real-time user behavior data matrix is ​​sorted according to timestamps to form a time-series dataset; the data points at each time step include all n labeled regions. and data; Set the time window length to T time steps, and divide the time series data into multiple input samples. Each sample contains continuous time step data of length T, which are used as input to the LSTM model. ； in, These are training samples, which are input sequences consisting of data from consecutive time steps, used as input to the LSTM model. It is the sample index, indicating the starting position of the currently generated training sample in the time series; M represents the total number of time steps in the time series dataset. The preprocessed time series dataset was divided into training, validation and test sets in an 80:10:10 ratio to ensure that the model has good generalization ability. The latest user behavior data matrix at time step T Input the trained Bi-LSTM model; The model calculates and generates the predicted behavior weight matrix. ,in This represents the predicted weight value for the i-th labeled region; This indicates the importance of the interaction of the i-th marked region in the next time step; a higher value indicates a higher priority. The predicted behavior weight matrix Output in a structured data format, including the weight values ​​of each labeled region. .

5. The dynamic backlight compensation optimization method for multi-user screen interaction according to claim 4, characterized in that: The initial backlight brightness adjustment matrix is ​​used to initially control the luminous intensity of the LEDs in each marked area on the target screen; Brightness changes are achieved by adjusting the electrical signal strength of the LEDs through the screen control unit; Let high priority, medium priority, and low priority be denoted as G1, G2, and G3, respectively; The adjustable range of backlight brightness for each marked area on the target screen is set to [Ld1, Ld2], in % and ; Assign a corresponding relative brightness ratio coefficient based on the priority of each marked region j. : ； in, This is an abbreviation for the priority level of the marked region j; When the marked region j is of high priority, the relative brightness ratio coefficient is... Characterized as 1.0; This indicates that the initial backlight brightness of the marked area j is set to the maximum brightness of the adjustable backlight brightness range [Ld1, Ld2], which is Ld2. When the marked region j is of medium priority, the relative brightness ratio coefficient will be... The value is 0.5; This indicates that the initial backlight brightness of the marked area j is set to half of the maximum brightness within the adjustable backlight brightness range [Ld1, Ld2]. ; When the marked region j is of low priority, the relative brightness ratio coefficient is... The value is 0.2; This indicates that the initial backlight brightness of the marked area j is set to the lowest adjustable brightness within the adjustable backlight brightness range [Ld1, Ld2], which is Ld1. set up The value range is [0,1]; relative brightness ratio coefficient Used to represent the ratio of the initial backlight brightness to Ld2; Let the initial backlight brightness adjustment matrix be denoted as .

6. The dynamic backlight compensation optimization method for multi-user screen interaction according to claim 5, characterized in that: Adjust the matrix based on the initial backlight brightness. Calculate the energy consumption of each marked area; ； in, It is the reference energy consumption of the light-emitting diode in the marked area j under the unit relative brightness ratio coefficient; This is the actual energy consumption of the light-emitting diode in the marked area j; Calculate the total energy consumption of the target screen. Set the preset adjustment threshold to the adjustment threshold range. ; and Total energy consumption Adjust the lower and upper limits of the threshold; like Keep the initial backlight brightness value unchanged; like Then reduce the initial backlight brightness value of the medium-priority and high-priority areas to make Back ; like This increases the initial backlight brightness value of the medium-priority and high-priority areas, so that... Raise to Location; The adjusted relative brightness ratio coefficient of the marked region j is denoted as... ; The final backlight brightness adjustment matrix is ​​characterized as .

7. The dynamic backlight compensation optimization method for multi-user screen interaction according to claim 6, characterized in that: The final backlight brightness adjustment matrix is ​​characterized as Converted into an electrical signal strength format that the screen control unit can recognize; The input terminal of the screen control unit receives the final backlight brightness adjustment matrix, which is then transmitted to the control unit via an electrical signal strength format. After analysis and processing, the control unit generates a corresponding LED light intensity adjustment command based on the received electrical signal strength, so as to dynamically adjust the LED light intensity of each marked area on the target screen.

Images