A dynamic response adjustment method of a touch screen

Abstract

The application relates to the technical field of touch screen control, in particular to a dynamic response adjustment method for a touch screen, which comprises the following steps: first, obtaining interface elements of the touch screen; then, obtaining a primary touch heat map according to the interface elements, and adjusting the dynamic response of the touch screen based on the primary touch heat map; next, obtaining screen pixels and touch records in the past period of time; then, predicting a predicted touch heat map according to the screen pixels and the touch records based on a preset neural network model; finally, fusing the primary touch heat map and the predicted touch heat map to obtain a real-time touch heat map, and adjusting the dynamic response of the touch screen based on the real-time touch heat map. Through the intelligent adjustment mechanism driven by real-time data, the dynamic adjustment of the touch screen parameters according to real-time conditions and user behaviors is realized, and the problem that the touch screen cannot dynamically respond in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of touch screen control, and in particular to a dynamic response adjustment method for a touch screen. BACKGROUND

[0002] In today's digital interaction scenarios, touch screens have become the core carrier of human-computer interaction, widely covering the fields of smart phones, tablet computers, public information terminals and industrial touch devices. Its importance is reflected in efficient and convenient operation experience, optimized structure design without physical keys, and integrated capabilities of multiple types of sensing technology.

[0003] However, most existing touch screen systems adopt a fixed response strategy, i.e., preset constant threshold parameters are used to process touch behavior, resulting in adaptation limitations in different use scenarios. When the user has no touch intention, the interface may trigger a false touch due to sensitive response, and when fine operation is required, the threshold may be too high to cause delayed feedback.

[0004] Therefore, there is an urgent need for an intelligent response mechanism that can dynamically adjust touch screen parameters according to real-time conditions and user behavior to break through the limitations of traditional solutions and improve user experience while ensuring accuracy. SUMMARY

[0005] Therefore, the present application provides a dynamic response adjustment method for a touch screen to solve the problem that the touch screen cannot dynamically respond in the prior art.

[0006] The present application provides a dynamic response adjustment method for a touch screen, comprising:

[0007] obtaining interface elements of the touch screen;

[0008] obtaining a preliminary touch heat map based on the interface elements, and dynamically adjusting the touch screen based on the preliminary touch heat map;

[0009] obtaining screen pixels and touch records in the past period of time;

[0010] predicting a predicted touch heat map based on the screen pixels and the touch records based on a preset neural network model;

[0011] fusing the preliminary touch heat map and the predicted touch heat map to obtain a real-time touch heat map, and dynamically adjusting the touch screen based on the real-time touch heat map;

[0012] The preliminary touch heat map, the predicted touch heat map and the real-time touch heat map are all used to represent the probability of the user clicking different positions on the touch screen.

[0013] In one preferred scheme: obtaining a primary touch heat map according to the interface element, and adjusting the dynamic response of the touch screen based on the primary touch heat map, including:

[0014] Obtaining application information of the current display content of the touch screen;

[0015] According to the application information, obtaining a preset touch heat map corresponding to the application information, wherein the preset touch heat map is used to represent the probability of user clicking different positions on the touch screen;

[0016] According to the interface element, obtaining the touch button position;

[0017] According to the touch button position, correcting the preset touch heat map to obtain a primary touch heat map;

[0018] Adjusting the dynamic response of the touch screen based on the primary touch heat map.

[0019] In one preferred scheme: according to the touch button position, correcting the preset touch heat map to obtain a primary touch heat map, including:

[0020] Dividing a grid in the preset touch heat map, and according to the touch button position, counting whether there is a touch button in each grid;

[0021] If there is a touch button in the target grid, then the touch probability represented by all grids in the preset neighborhood centered on the target grid in the preset touch heat map is increased to obtain a primary touch heat map; wherein the touch probability of all grids in the preset neighborhood decreases from the center to the edge.

[0022] In one preferred scheme: based on a preset neural network model, a predicted touch heat map is predicted according to the screen pixels and the touch record, including:

[0023] Dividing the screen area into a plurality of pixel blocks;

[0024] According to the distribution of screen pixels in each pixel block, establishing a pixel feature;

[0025] According to the distribution of touch records in each pixel block, establishing a touch feature;

[0026] Inputting the pixel feature and the touch feature into the preset neural network model to obtain the predicted touch heat map output by the preset neural network.

[0027] In one preferred scheme: according to the distribution of screen pixels in each pixel block, establishing a pixel feature, including:

[0028] According to the screen pixels, counting the average value of H / S / V channels in HSV space of the pixels in each pixel block to obtain a color histogram;

[0029] According to the screen pixels, the local binary pattern mean value of the gray image in each pixel block is counted to obtain a texture feature value;

[0030] According to the screen pixels, the difference between the mean value of the pixel gray value in the current frame and the mean value of the pixel gray value in the previous frame is counted to obtain a dynamic feature value;

[0031] A vector is established according to the color histogram, the texture feature value and the dynamic feature value as the pixel feature.

[0032] In a preferred scheme: the touch record includes touch position coordinates, operation type and contact area; according to the distribution of the touch record in each pixel block, a touch feature is established;

[0033] The touch position coordinates are mapped into the index of the pixel block where the touch position coordinates are located to obtain a position feature;

[0034] The operation type is encoded to obtain an operation feature;

[0035] The contact area is normalized to obtain a contact feature;

[0036] A vector is established according to the position feature, the operation feature and the contact feature as the touch feature.

[0037] In a preferred scheme: the preset neural network model includes a CNN neural network layer, an LSTM neural network layer, a feature fusion layer and an output layer, wherein the CNN neural network layer includes a first input layer, a one-dimensional convolution layer, a first pooling layer and a first output layer connected in sequence, wherein the first input layer is used for inputting the pixel feature, the one-dimensional convolution layer includes a plurality of Conv1d convolution kernels, the number of Conv1d convolution kernels is the same as the number of pixel blocks, the first pooling layer is used for dimension reduction on the output vector of the one-dimensional convolution layer, and the first output layer is used for outputting a first context vector; the LSTM neural network layer includes a second input layer, an LSTM layer, a second pooling layer and a second output layer connected in sequence, wherein the second input layer is used for inputting the touch feature multiple times, the LSTM neural network layer includes a plurality of LSTM units, the second pooling layer is used for dimension reduction on the output vector of the LSTM neural network layer, and the second output layer is used for outputting a second context vector; the feature fusion layer includes a vector splicing layer, a first full connection layer, a second full connection layer and a third output layer connected in sequence, wherein the vector splicing layer is used for splicing the first context vector and the second context vector, the first full connection layer is used for dimension reduction and analysis of the feature, the second full connection layer is used for outputting the heat weight of each pixel block, and the third output layer is used for outputting the predicted touch heat map.

[0038] In a preferred scheme: the dynamic response adjustment method of the touch screen further includes:

[0039] The preset touch heatmap is corrected based on the real-time touch heatmap.

[0040] The present invention also provides an electronic device, comprising:

[0041] Memory and processor;

[0042] The memory is used to store the program, and the processor is used to execute the steps in any of the above-described methods for adjusting the dynamic response of a touch screen when the program is executed.

[0043] The present invention also provides a computer-readable storage medium for storing a computer-readable program or instruction, which, when executed by a processor, can implement the steps in any of the above-described dynamic response adjustment methods for a touch screen.

[0044] The beneficial effects of using the above embodiments are:

[0045] This invention provides a method for dynamically adjusting the response of a touch screen. First, it acquires the interface elements of the touch screen, then obtains a primary touch heatmap based on these elements, and dynamically adjusts the touch screen's response based on this heatmap. Next, it acquires screen pixels and touch records from a past period, and predicts a touch heatmap based on a preset neural network model. Finally, it merges the primary and predicted touch heatmaps to obtain a real-time touch heatmap, and dynamically adjusts the touch screen's response based on this real-time heatmap. This invention first generates a primary touch heatmap based on interface elements, quickly responding to the current interface layout features. Then, combining screen pixel states and historical touch records, it predicts potential user operation areas using a preset neural network model, generating a predictive touch heatmap for proactive response. By merging the two types of heatmaps into a real-time touch heatmap, it dynamically adjusts the touchscreen's sensitivity and response parameters, avoiding accidental touches or delays caused by fixed thresholds and adapting to different operating habits and environmental changes in various scenarios. This invention achieves dynamic adjustment of touchscreen parameters based on real-time conditions and user behavior through a real-time data-driven intelligent adjustment mechanism, thus solving the problem of touchscreens being unable to respond dynamically in the prior art. Attached Figure Description

[0046] Figure 1 A flowchart of the dynamic response adjustment method for a touch screen provided by the present invention;

[0047] Figure 2 for Figure 1 A detailed step diagram of step S102 is shown below;

[0048] Figure 3 for Figure 1 The detailed steps of step S104 are shown in the diagram. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0050] In combination with Figure 1 As shown in FIG. 1, one specific embodiment of the present application discloses a dynamic response adjustment method of a touch screen, comprising the following steps:

[0051] S101, obtaining interface elements of the touch screen;

[0052] S102, obtaining a primary touch heat map according to the interface elements, and performing dynamic response adjustment on the touch screen based on the primary touch heat map;

[0053] S103, obtaining screen pixels and touch records in a past period of time;

[0054] S104, predicting a predicted touch heat map according to the screen pixels and the touch records based on a preset neural network model;

[0055] S105, fusing the primary touch heat map and the predicted touch heat map to obtain a real-time touch heat map, and performing dynamic response adjustment on the touch screen based on the real-time touch heat map;

[0056] The primary touch heat map, the predicted touch heat map and the real-time touch heat map are all used to represent the probability of a user clicking different positions on the touch screen.

[0057] The primary touch heat map, the predicted touch heat map, the real-time touch heat map and the preset touch heat map to be mentioned later in the above three processes have the same meaning but different functions. The primary touch heat map is mainly used to quickly realize dynamic response adjustment and guarantee the basic use experience of the user. The predicted touch heat map is used for analysis of the real-time use state. The real-time touch heat map is used for fine response adjustment (the specific fusion method can be flexibly designed according to specific conditions, for example, taking the maximum value or weighted summation while limiting fluctuation). The preset touch heat map in the later text is a pre-set touch heat map of different applications.

[0058] The application firstly generates a primary touch heat map based on interface elements, quickly responds to the current interface layout characteristics, and then combines the screen pixel state and historical touch records to predict the user's potential operation area through a preset neural network model, generate a predicted touch heat map, realize forward-looking response, and fuse the two types of heat maps into a real-time touch heat map, dynamically adjust the touch screen sensitivity and response parameters, which not only avoids the false touch or delay problem caused by the fixed threshold, but also adapts to the operation habits and environmental changes in different scenes. Through the intelligent adjustment mechanism driven by real-time data, the application realizes dynamic adjustment of touch screen parameters according to real-time conditions and user behavior, and solves the problem that the touch screen cannot dynamically respond in the prior art.

[0059] In the above process, the dynamic response adjustment can be any existing response adjustment mode such as increasing TouchSlop (sliding trigger distance threshold), increasing the minimum pressing time of the click event (MIN_CLICK_DELAY), expanding the touch effective area (such as requiring the touch point area to be greater than a threshold), keeping the system default parameters or reducing the threshold, etc. It can be determined according to the actual situation. For example, when the user watches a full-screen video for a long time, the sensitivity of the edge area of the screen can be reduced to avoid false touch between the user's finger and the screen when holding the screen.

[0060] Further, in combination with Figure 2 It is shown that in a preferred embodiment, the above step S102, obtaining a primary touch heat map according to interface elements, and dynamically adjusting the touch screen based on the primary touch heat map, specifically includes:

[0061] S201, obtaining application information of the current display content of the touch screen;

[0062] S202, obtaining a preset touch heat map corresponding to the application information according to the application information, wherein the preset touch heat map is used to represent the probability of the user clicking different positions on the touch screen;

[0063] S203, obtaining a touch button position according to the interface elements;

[0064] S204, correcting the preset touch heat map according to the touch button position to obtain a primary touch heat map;

[0065] S205, dynamically adjusting the touch screen based on the primary touch heat map.

[0066] In the above process, the current foreground application package name can be obtained as application information through a system API (such as ActivityManager of Android) to determine whether it is a full-screen application such as a video player or a game. An accessibility service (such as AccessibilityService of Android) can be used to traverse the current interface view tree to detect full-screen controls (such as video player controls) or split-screen regions to obtain interface elements.

[0067] The embodiment first calls a preset touch heat map template according to current application information, quickly establishes a basic response model using application scenario characteristics (such as games, reading, drawing, etc.), locates the positions of key controls such as touch buttons through interface element analysis, corrects the heat map distribution accordingly, eliminates the generalization error of the preset template, ensures high-precision matching of actual interaction requirements, and enables the system to quickly respond to high-frequency operation areas (such as navigation bars) and accurately identify low-frequency but key control interactions (such as hidden function buttons).

[0068] Further, in a preferred embodiment, the step S204 of correcting the preset touch heat map according to the touch button position to obtain a preliminary touch heat map specifically includes:

[0069] dividing a grid in the preset touch heat map, and counting whether there is a touch button in each grid according to the touch button position;

[0070] if there is a touch button in the target grid, increasing the touch probability represented by all grids in a preset neighborhood centered on the target grid in the preset touch heat map to obtain a preliminary touch heat map; wherein the touch probability of all grids in the preset neighborhood decreases from the center to the edge.

[0071] The embodiment significantly improves the accuracy and edge adaptability of the touch heat map through the grid-based probability enhancement mechanism. When a target grid is detected to contain a touch button, not only the touch probability of the grid is increased, but also the probability enhancement effect is smoothly transitioned to surrounding grids through neighborhood diffusion, forming a gradient probability distribution centered on the button. This design not only highlights the high-response priority of the core interaction area, but also avoids the response mutation problem caused by traditional hard boundary correction through gradual increase of the edge probability.

[0072] Further, in combination with Figure 3 In a preferred embodiment, the step S104 of predicting a predicted touch heat map based on a preset neural network model according to screen pixels and touch records specifically includes:

[0073] S301, dividing a screen area into a plurality of pixel blocks;

[0074] S302, establish pixel features according to the distribution of screen pixels in each pixel block;

[0075] S303, establish touch features according to the distribution of touch records in each pixel block;

[0076] S304, input the pixel features and the touch features into a preset neural network model to obtain a predicted touch heat map output by the preset neural network.

[0077] The embodiment divides the screen into a plurality of pixel blocks, respectively extracts screen pixel distribution features (such as luminance, color, texture, and visual information) and touch record distribution features (such as historical click frequency, pressure distribution, and interactive data) of each pixel block, and comprehensively captures screen states and user behavior patterns through double-channel feature modeling. Based on the multi-modal fusion capability of the deep neural network, the scheme realizes accurate prediction of the user's potential touch intention, significantly improves the intelligent level of dynamic response adjustment. More importantly, the pixel block division mechanism replaces global screen data acquisition through local feature extraction, effectively avoids direct exposure of the original screen image while ensuring prediction accuracy, and fundamentally eliminates the risk of privacy leakage, providing a safe and reliable technical solution for touch interaction in sensitive scenarios such as mobile devices and public terminals.

[0078] Specifically, in one preferred embodiment, the step S302 of establishing pixel features according to the distribution of screen pixels in each pixel block specifically includes:

[0079] According to the screen pixels, the mean values of H / S / V channels of the pixels in each pixel block in the HSV space are counted to obtain a color histogram;

[0080] According to the screen pixels, the local binary pattern mean value of the grayscale image in each pixel block is counted to obtain a texture feature value;

[0081] According to the screen pixels, the difference between the mean value of the pixel grayscale value in the current frame and the mean value of the pixel grayscale value in the previous frame in each pixel block is counted to obtain a dynamic feature value;

[0082] A vector is established according to the color histogram, the texture feature value, and the dynamic feature value as the pixel features.

[0083] In the above process, the mean value information of hue (H), saturation (S) and lightness (V) in the HSV color space is decomposed to construct a color histogram, which can effectively capture the color distribution characteristics of the interface elements. The texture feature value is counted by the local binary pattern (LBP), which accurately describes the microscopic structure of the screen surface. The dynamic feature is extracted by calculating the gray difference value between the current frame and the previous frame, which can realize real-time perception of the interface change trend. The three-dimensional feature vector integrating color, texture and dynamic change not only comprehensively represents the visual characteristics of the pixel block, but also enhances the environmental adaptability of touch prediction through time sequence correlation analysis.

[0084] Specifically, in a preferred embodiment, the touch record includes touch position coordinates, operation type and contact area, on the basis of which, the step S303 of establishing touch features according to the distribution of the touch record in each pixel block specifically includes:

[0085] Mapping the touch position coordinates to the index of the pixel block where the touch position coordinates are located to obtain the position feature;

[0086] Encoding the operation type to obtain the operation feature;

[0087] Normalizing the contact area to obtain the contact feature;

[0088] Establishing a vector according to the position feature, the operation feature and the contact feature as the touch feature.

[0089] In this embodiment, the position coordinates in the touch record are converted into pixel block indexes to construct a discretized position feature, which not only retains the spatial distribution information but also reduces the data dimension. The three-dimensional feature vector integrating spatial position, operation semantics and physical interaction intensity in this embodiment completely describes the essential features of the touch event, so that the neural network can more accurately learn the user behavior pattern. It can be understood that the touch feature is a sequence of multiple vectors.

[0090] Specifically, in a preferred embodiment, the preset neural network model comprises a CNN neural network layer, an LSTM neural network layer, a feature fusion layer and an output layer, wherein the CNN neural network layer comprises a first input layer, a one-dimensional convolution layer, a first pooling layer and a first output layer connected in sequence, wherein the first input layer is used for inputting pixel features, the one-dimensional convolution layer comprises a plurality of Conv1d convolution kernels, the number of Conv1d convolution kernels is the same as the number of pixel blocks, the first pooling layer is used for dimension reduction on the output vector of the one-dimensional convolution layer, and the first output layer is used for outputting a first context vector; the LSTM neural network layer comprises a second input layer, an LSTM layer, a second pooling layer and a second output layer connected in sequence, wherein the second input layer is used for inputting touch features multiple times, the LSTM neural network layer comprises a plurality of LSTM units, the second pooling layer is used for dimension reduction on the output vector of the LSTM neural network layer, and the second output layer is used for outputting a second context vector; the feature fusion layer comprises a vector splicing layer, a first full connection layer, a second full connection layer and a third output layer connected in sequence, wherein the vector splicing layer is used for splicing the first context vector and the second context vector, the first full connection layer is used for dimension reduction and analysis of features, the second full connection layer is used for outputting a heat weight of each pixel block, and the third output layer is used for outputting a predicted touch heat map.

[0091] The application also provides a more specific embodiment for illustrating the above design.

[0092] In this embodiment, the screen is divided into 8x8 pixel blocks (a total of 64 blocks), and the touch record is a touch event sequence in the last 5 seconds (a maximum of 50). The pixel features extracted from each block include:

[0093] Color histogram: mean value of H / S / V channels in HSV space (3 dimensions).

[0094] Texture feature value: local binary pattern (LBP) mean value of the gray image (1 dimension).

[0095] Dynamic feature value: average pixel difference between the current frame and the previous frame of the gray image (1 dimension).

[0096] The vector dimension of the constructed pixel features: 64 blocks x (3+1+1)=320 dimensions.

[0097] The touch features extracted from each block include:

[0098] Position feature: coordinate mapping to the corresponding pixel block index (for example, x=100, y=200 is mapped to block_idx=15).

[0099] Operation feature: One-hot encoding for operation types (click / slide / long press) (3 dimensions).

[0100] Contact feature: fingertip contact area (normalized to 0~1).

[0101] Sequence padding: zero padding when less than 50, truncation when too long, keeping the latest 50.

[0102] The vector dimension of the constructed touch feature is (1+3+1)=5, but there are a total of 50.

[0103] The preset neural network design takes lightweight as the core, adopts a double-branch parallel processing structure, is divided into a CNN branch (i.e., a CNN neural network, processing pixel features) and an LSTM branch (i.e., an LSTM neural network, processing touch feature sequences), and finally outputs 64 pixel block heat weights through feature fusion.

[0104] The CNN neural network layer includes a first input layer, a one-dimensional convolution layer, a first pooling layer, and a first output layer connected in sequence.

[0105] In the first input layer, the input channel number is 64 blocks (each block has a 5-dimensional feature). The one-dimensional convolution layer is used to extract local spatial correlation (such as color / texture difference between adjacent pixel blocks). In the one-dimensional convolution layer, Conv1d is used instead of traditional CNN to reduce spatial calculation, a small kernel (3x1) is used to reduce the number of parameters, no bias term is set to further compress parameters, 64 convolution kernels are set, each kernel generates one feature map, and finally 64 blocks of 64-dimensional features are output. In the first pooling layer, the dimension is reduced to 16 blocks to reduce the amount of subsequent calculation. The 64 blocks of features are compressed into 16 blocks, and the mean value of each adjacent 4 blocks is taken. Finally, the first output layer outputs a 1024-dimensional first context vector. The CNN neural network layer adopts a ReLU activation function to accelerate convergence and suppress gradient disappearance.

[0106] The LSTM neural network layer includes a second input layer, an LSTM layer, a second pooling layer, and a second output layer connected in sequence.

[0107] The second input layer is used to input the pixel feature with a dimension of 5 multiple times. In the LSTM layer, a Tanh activation function is used to process the internal control signal of LSTM, and the value range is constrained in [-1, 1]. A Sigmoid activation function is used to control the gate weight of information forgetting / transferring. The second pooling layer is used to retain the most significant operation mode (such as high-frequency click area), that is, only the output of the last 8 time steps is retained to avoid processing the full sequence. Finally, the second output layer outputs a 128-dimensional second context vector.

[0108] The feature fusion layer includes a vector splicing layer, a first full connection layer, a second full connection layer, and a third output layer connected in sequence.

[0109] The vector splicing layer is used for splicing the first context vector and the second context vector, and outputs a 1024+128=1152-dimensional vector, the first full connection layer is used for dimension reduction and feature analysis, and data is reduced to 32 dimensions, cross-modal feature interaction is learned, and a ReLU activation function is used to accelerate convergence, the second full connection layer is used for further analysis to output the heat weight of 64 pixel blocks, and the third output layer is used for outputting the predicted touch heat map, and a Sigmoid activation function is used to realize normalization, that is, the heat weight corresponding to each pixel block is limited between 0 and 1, representing a probability weight.

[0110] This design realizes adaptive touch sensitivity adjustment by using a light network structure and feature engineering to accurately capture user behavior and pixel dynamic characteristics while ensuring real-time performance.

[0111] Further, in a preferred embodiment, the dynamic response adjustment method of the touch screen further comprises:

[0112] According to the real-time touch heat map, the preset touch heat map is corrected.

[0113] It should be noted that the above steps in the present application and the steps S103-S105 in the foregoing can be performed in the cloud, and a certain trigger condition (such as long-time video watching or game software calling) can be set to avoid affecting local computing power. The present embodiment further proposes a feedback mechanism, which dynamically corrects the preset touch heat map through the real-time touch heat map, and significantly improves the long-term adaptability and user experience of the system.

[0114] The present application also provides an electronic device, characterized in that it comprises:

[0115] a memory and a processor;

[0116] The memory is used for storing a program, and the processor is used for executing the steps of the dynamic response adjustment method of the touch screen in any of the above embodiments when executing the program.

[0117] The present application also provides a computer readable storage medium for storing computer readable programs or instructions, which can realize the steps of the dynamic response adjustment method of the touch screen in any of the above embodiments when the programs or instructions are executed by a processor.

[0118] The application provides a dynamic response adjustment method of a touch screen, which comprises the following steps: obtaining interface elements of the touch screen, obtaining a primary touch heat map according to the interface elements, adjusting the dynamic response of the touch screen based on the primary touch heat map, obtaining screen pixels and touch records in the past period of time, predicting a predicted touch heat map based on a preset neural network model according to the screen pixels and the touch records, fusing the primary touch heat map and the predicted touch heat map to obtain a real-time touch heat map, and adjusting the dynamic response of the touch screen based on the real-time touch heat map. The primary touch heat map is generated based on the interface elements, the current interface layout features are quickly responded, the potential operation area of the user is predicted by the preset neural network model by combining the screen pixel state and the historical touch records, the predicted touch heat map is generated, the prospective response is realized, the two types of heat maps are fused into the real-time touch heat map, the sensitivity and the response parameters of the touch screen are dynamically adjusted, the false touch or the delay problem caused by the fixed threshold is avoided, and the operation habits and the environmental changes in different scenes can be adapted. The intelligent adjustment mechanism driven by the real-time data is used, the touch screen parameters are dynamically adjusted according to the real-time situation and the user behavior, and the problem that the touch screen cannot dynamically respond in the prior art is solved.

[0119] It should be noted that each of the embodiments in the specification adopts a progressive manner for description, and each embodiment focuses on the difference from other embodiments, and the same and similar parts between the embodiments can be referred to each other.

[0120] The above description of the disclosed embodiments enables a person skilled in the art to implement or use the application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the application. Therefore, the application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for adjusting the dynamic response of a touch screen, characterized in that, include: Get the UI elements of the touch screen; A preliminary touch heatmap is obtained based on the interface elements, and the touch screen is dynamically adjusted based on the preliminary touch heatmap. Obtain screen pixels and touch records over a past period of time; Based on a preset neural network model, a predictive touch heatmap is generated according to screen pixels and touch records. By fusing the primary touch heatmap and the predicted touch heatmap, a real-time touch heatmap is obtained. Based on the real-time touch heatmap, the touch screen is dynamically adjusted to regulate the sensitivity of the touch screen in response to user touches. Among them, the primary touch heatmap, the predictive touch heatmap, and the real-time touch heatmap are all used to represent the probability of a user clicking on different locations on the touch screen.

2. The dynamic response adjustment method for a touch screen according to claim 1, characterized in that, A preliminary touch heatmap is obtained based on the interface elements, and the touch screen's dynamic response is adjusted based on this preliminary touch heatmap, including: Get application information about the content currently displayed on the touchscreen; Based on the application information, obtain the preset touch heatmap corresponding to the application information. The preset touch heatmap is used to represent the probability of a user clicking on different locations on the touch screen. Based on the interface elements, determine the location of the touch button; Based on the location of the touch buttons, the preset touch heat map is corrected to obtain a primary touch heat map; Dynamic response adjustment of the touch screen is performed based on a primary touch heatmap.

3. The dynamic response adjustment method for a touch screen according to claim 2, characterized in that, Based on the location of the touch buttons, the preset touch heatmap is corrected to obtain a primary touch heatmap, including: Divide the preset touch heatmap into a grid, and count whether there is a touch button in each grid according to the position of the touch button; If a touch button exists within the target grid, the touch probability represented by all grids in the preset neighborhood centered on the target grid in the preset touch heatmap is increased to obtain a primary touch heatmap; wherein, the increase in touch probability of all grids in the preset neighborhood decreases sequentially from the center to the edge.

4. The dynamic response adjustment method for a touch screen according to claim 2, characterized in that, Based on a pre-defined neural network model, a predicted touch heatmap is generated according to screen pixels and touch records, including: Divide the screen area into multiple pixel blocks; Pixel features are established based on the distribution of screen pixels within each pixel block; Based on the distribution of touch records within each pixel block, touch features are established; Pixel features and touch features are input into a preset neural network model to obtain a predicted touch heatmap output by the preset neural network.

5. The dynamic response adjustment method for a touch screen according to claim 4, characterized in that, Based on the distribution of screen pixels within each pixel block, pixel features are established, including: Based on the screen pixels, the mean values ​​of the H / S / V channels of the pixels within each pixel block in the HSV space are calculated to obtain the color histogram; Based on the screen pixels, the mean value of the local binary pattern of the grayscale image within each pixel block is calculated to obtain the texture feature value; Based on the screen pixels, the difference between the average grayscale value of each pixel block in the current frame and the average grayscale value in the previous frame is calculated to obtain the dynamic feature value. Vectors are created based on color histograms, texture features, and dynamic features to serve as pixel features.

6. The dynamic response adjustment method for a touch screen according to claim 4, characterized in that, Touch records include touch location coordinates, operation type, and contact area; touch features are established based on the distribution of touch records within each pixel block. The touch position coordinates are mapped to the index of the pixel block in which the touch position coordinates are located, thus obtaining the position features; The operation type is encoded to obtain the operation characteristics; The contact area is normalized to obtain the contact characteristics; Vectors are created based on position features, operation features, and contact features, serving as touch features.

7. The dynamic response adjustment method for a touch screen according to claim 4, characterized in that, The pre-defined neural network model includes a CNN neural network layer, an LSTM neural network layer, a feature fusion layer, and an output layer. The CNN neural network layer comprises a first input layer, a one-dimensional convolutional layer, a first pooling layer, and a first output layer, all connected in sequence. The first input layer is used to input pixel features; the one-dimensional convolutional layer includes multiple Conv1d convolutional kernels, the number of which is the same as the number of pixel blocks; the first pooling layer is used to reduce the dimensionality of the output vector of the one-dimensional convolutional layer; and the first output layer outputs a first context vector. The LSTM neural network layer comprises a second input layer, an LSTM layer, and a second pooling layer, all connected in sequence. The system consists of a first context layer and a second output layer. The second input layer is used to input touch features multiple times. The LSTM neural network layer includes multiple LSTM units. The second pooling layer is used to reduce the dimensionality of the output vector of the LSTM neural network layer. The second output layer is used to output the second context vector. The feature fusion layer includes a vector concatenation layer, a first fully connected layer, a second fully connected layer, and a third output layer connected in sequence. The vector concatenation layer is used to concatenate the first context vector and the second context vector. The first fully connected layer is used to reduce the dimensionality and analyze the features. The second fully connected layer is used to output the heat weight of each pixel block. The third output layer is used to output the predicted touch heatmap.

8. The dynamic response adjustment method for a touch screen according to claim 2, characterized in that, The dynamic response adjustment method for the touch screen further includes: The preset touch heatmap is corrected based on the real-time touch heatmap.

9. An electronic device, characterized in that, include: Memory and processor; The memory is used to store a program, and the processor is used to execute the steps of the dynamic response adjustment method for a touch screen according to any one of claims 1-8 when executing the program.

10. A computer-readable storage medium, characterized in that, Used to store computer-readable programs or instructions, which, when executed by a processor, enable the implementation of the steps in the dynamic response adjustment method for a touch screen according to any one of claims 1-8.

Images