Method for establishing a wet mode touch recognition model based on deep learning

Abstract

The application discloses a wet mode touch recognition model establishment method based on deep learning, relates to the technical field of wet touch recognition, and comprises the following steps: establishing a wet mode touch recognition model, configuring test samples and collecting test data; collecting features of the test data, acquiring spatial features and time features of finger touch in a wet environment; performing deep learning on the spatial features and the time features through the wet mode touch recognition model, and analyzing change rules of the spatial features and the time features; and identifying finger touch of a user in a wet mode through the wet mode touch recognition model based on the change rules; and the application is used for solving the problem that existing wet touch recognition technology cannot accurately identify the capacitive signals generated when finger touch in a wet area, which causes the touch screen to be prone to false touch.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wet touch control recognition, in particular to a wet mode touch control recognition model establishment method based on deep learning. BACKGROUND

[0002] Wet touch control recognition technology refers to a touch control technology specially designed for high humidity environments or liquid interference scenarios, aiming to solve the problems of false touch, drift or failure of traditional capacitive touch screens caused by wet factors such as water droplets, wet hands and water mist. It mainly optimizes hardware design and cooperates with intelligent algorithms to accurately distinguish between real touch intention and liquid noise signals.

[0003] The existing wet touch control recognition technology usually uses a preset threshold to determine the user's finger touch area in the wet mode, but the accuracy of the preset threshold is low, the false touch rate is high, and the liquid also generates a capacitive signal on the screen surface. When the finger touches the wet area, it will be affected by the liquid, and the change amplitude of the capacitive signal generated by the finger when touching the wet area is very small, so it is necessary to calculate the capacitive threshold more accurately when touching. For example, in the patent application with publication number CN114610176A, a "screen control method, device and computer readable storage medium" is disclosed. The scheme judges by a preset threshold when recognizing touch in the wet mode, which has low accuracy and is prone to false touch. The existing wet touch control recognition technology also has the problem of being difficult to accurately recognize the capacitive signal generated by the finger when touching the wet area, which causes the touch screen to be prone to false touch. SUMMARY

[0004] The present application aims to at least solve one of the technical problems in the prior art by establishing a wet mode touch control recognition model, configuring test samples and collecting test data, then collecting features of CC(n, i, j) in the test data to obtain spatial features of finger touch in a wet environment, collecting features of the test data as a whole to obtain time features of finger touch in a wet environment, then performing deep learning on the spatial features through the wet mode touch control recognition model to analyze the spatial variation law of the spatial features, and performing deep learning on the time features through the wet mode touch control recognition model to analyze the time variation law of the time features. Finally, based on the variation law, the wet mode touch control recognition model is used to recognize the user's finger touch in the wet mode, to solve the problem that the existing wet touch control recognition technology is difficult to accurately recognize the capacitive signal generated by the finger when touching the wet area, which causes the touch screen to be prone to false touch.

[0005] To achieve the above purpose, the present application provides a wet mode touch control recognition model establishment method based on deep learning, comprising the following steps:

[0006] Establish a touch recognition model for a humid mode, configure test samples, and collect test data;

[0007] Feature acquisition was performed on the test data to obtain the spatial and temporal features of finger touch in a humid environment;

[0008] By using a wet-mode touch recognition model, deep learning is performed on spatial and temporal features to analyze the changing patterns of spatial and temporal features.

[0009] Based on the changing patterns, a wet-mode touch recognition model is used to identify users' finger touches in wet conditions.

[0010] Furthermore, establishing a touch recognition model for a humid environment, configuring test samples, and collecting test data includes the following sub-steps:

[0011] A touch recognition model for wet conditions was built using a convolutional neural network.

[0012] A capacitive sensing matrix is ​​installed behind the touchscreen to monitor the capacitive signals on the touchscreen.

[0013] Construct a humid environment group and a humid touch group, pour water stains on the touch screen in the humid environment group, pour water stains on the touch screen in the humid touch group, and touch the touch screen with your finger;

[0014] The capacitive sensors in the capacitive sensing matrix are numbered using the symbol CS. ij This represents the expression, where i and j are both positive integers and i and j are the indices of CS, and CS... ij This represents the capacitive sensor located in the i-th row and j-th column, arranged from top to bottom and from left to right.

[0015] CS ij The detected capacitance signal is marked as CG. ij Simultaneously recording CG ij The monitoring time is numbered and marked as T in order from morning to night. n Where n is a positive integer and n is the index of T, the index of T is... n CG ij Let CC(n,i,j) be the index of CC, and CC(n,i,j) represent belonging to T. n CG ij ;

[0016] Recording T in humid environments n And CC(n,i,j), respectively labeled as T1 n CC1(n,i,j), named the humid environment data, records the T of the humid touch group. nand CC(n, i, j), respectively, are marked as T2 n and CC2(n, i, j), named as wet touch data, the wet environment data and the wet touch data are test data.

[0017] Further, the test data is characterized to obtain the spatial and temporal characteristics of the finger touch in the wet environment, including the following sub-steps:

[0018] The CC(n, i, j) in the test data is characterized to obtain the spatial characteristics of the finger touch in the wet environment.

[0019] The test data is characterized as a whole to obtain the temporal characteristics of the finger touch in the wet environment.

[0020] Further, the CC(n, i, j) in the test data is characterized to obtain the spatial characteristics of the finger touch in the wet environment, including the following sub-steps:

[0021] Get CC1(n, i, j) and CC2(n, i, j), and analyze for the same value of n. A one-dimensional coordinate system is established with the capacitance signal as the horizontal axis, named as the capacitance distribution diagram, the capacitance signal is recorded in the capacitance distribution diagram, and the clustering analysis is performed on the capacitance distribution diagram to obtain the capacitance cluster.

[0022] Determine whether the number of capacitance clusters is two, if yes, output a clustering completion signal, if not, output a clustering incomplete signal, if the clustering incomplete signal is output, increase the clustering coefficient of the clustering analysis and perform the clustering analysis again until the number of capacitance clusters is two.

[0023] Calculate the average value of the capacitance signal in the capacitance cluster, named as the average signal, and the capacitance cluster with the maximum average signal is named as the sensitive cluster. CC1(n, i, j) and CC2(n, i, j) are recorded in the capacitance distribution diagram to analyze the sensitive cluster, and the sensitive cluster corresponding to CC1(n, i, j) and CC2(n, i, j) is named as the sensitive environment cluster and the sensitive touch cluster, respectively.

[0024] Calculate the average value of the capacitance signal in the sensitive environment cluster, named as the average capacitance, represented by the symbol ACE, and mark the capacitance signal in the sensitive touch cluster as STC(n, i, j). Calculate STC(n, i, j)-ACE, and name the calculation result as the capacitance difference.

[0025] Number the capacitance difference, and mark it as CD m , where m is a positive integer and m is the serial number of CD, the number of different capacitance differences is counted, and the number of CD m is marked as CQ m ;

[0026] CD m is taken as the X axis, and CQ m is taken as the Y axis to establish a two-dimensional coordinate system, which is named as a capacitance difference distribution graph, and CQ m is entered into the capacitance difference distribution graph, and the coordinate points in the capacitance difference distribution graph are named as capacitance difference distribution points; m

[0027] The adjacent capacitance difference distribution points are connected through a smooth curve, the finally connected curve is named as a capacitance difference distribution curve, the CD m corresponding to the lowest point of the capacitance difference distribution curve is obtained, and is named as a capacitance demarcation value;

[0028] CC2(n, i, j) less than the capacitance demarcation value is named as a capacitance environment signal, CC2(n, i, j) greater than or equal to the capacitance demarcation value is named as a capacitance touch signal, and the area where the capacitance touch signal corresponds to the capacitance sensor is named as a capacitance touch area, and the capacitance touch area is a spatial feature.

[0029] Further, the time feature of the finger touch in the humid environment is obtained by collecting the features of the test data as a whole, including the following sub-steps:

[0030] The average value of the capacitance signal in the capacitance touch area is calculated, and is named as a regional average capacitance;

[0031] The regional average capacitance of T n is marked as RAC n , a two-dimensional coordinate system is established with T n as the X axis and RAC n as the Y axis, which is named as a capacitance change graph, and RAC n is entered into the capacitance change graph according to T n ;

[0032] The capacitance change graph is linearly functionally fitted to obtain a capacitance change function, the slope of the capacitance change function is obtained, and is named as a capacitance change amplitude, and the capacitance change amplitude is the time feature.

[0033] Further, the spatial feature and the time feature are deeply learned through the humid mode touch recognition model, and the change law of the spatial feature and the time feature is analyzed, including the following sub-steps:

[0034] The spatial feature is deeply learned through the humid mode touch recognition model, and the spatial change law of the spatial feature is analyzed;

[0035] The time feature is deeply learned through the humid mode touch recognition model, and the time change law of the time feature is analyzed.

[0036] ​Further, the space features are learned in depth by the wet mode touch recognition model, and the space variation law of the space features is analyzed, including the following sub-steps:

[0037] All the space features in the wet touch data are collected, and the space features are numbered, and the symbol SF t is used, where t is a positive integer and t is the serial number of SF;

[0038] SF1 is named as a combination, t=2 is started, SF t is overlapped with the combination, and SF t is required to overlap, the overlapping area of SF t with the combination reaches the maximum, the combination is updated as the contour of the overlapped image, t+1 is added and overlapped again until the maximum value of t is reached;

[0039] The final combination is named as a space reference map, the similarity of SF t with the space reference map is calculated, and is marked as SY t .

[0040] The range of SY t is counted, and is named as a space variation law.

[0041] Further, the time features are learned in depth by the wet mode touch recognition model, and the time variation law of the time features is analyzed, specifically, the range of the time features is counted, that is, the time variation law.

[0042] Further, based on the variation law, the wet mode touch of the user's finger under the wet mode is recognized by the wet mode touch recognition model, including the following sub-steps:

[0043] The capacitance signals on the touch screen are monitored in real time, and the space features and the time features are analyzed, and are respectively named as space real-time features and time real-time features;

[0044] The similarity of the space real-time features with the space reference map is calculated, and is named as a space real-time similarity, if the space real-time similarity is within the space variation law, a space verification pass signal is output, otherwise a space verification fail signal is output;

[0045] It is judged whether the time real-time features are within the time variation law, if yes, a time verification pass signal is output, otherwise a time verification fail signal is output;

[0046] Based on the output signal, it is judged whether the touch under the wet mode is effective.

[0047] Further, based on the output signal, it is judged whether the touch under the wet mode is effective, including the following sub-steps:

[0048] If the space verification pass signal and the time verification pass signal are output simultaneously, it is marked that the touch in the wet mode is valid.

[0049] If the space verification fail signal or the time verification fail signal is output, it is marked that the touch in the wet mode is invalid.

[0050] The present application has the beneficial effects that: the present application establishes a wet mode touch recognition model, configures test samples and collects test data, then collects features of CC(n, i, j) in the test data, obtains spatial features of finger touch in a wet environment, and collects features of the test data as a whole, obtains time features of finger touch in a wet environment, the advantage is that the spatial features reveal the shape features when the finger touches in the wet state, and the difference between the capacitance signal generated by the liquid and the capacitance signal when touching in the liquid area is accurately analyzed in the analysis process, the capacitance dividing value is found in the weak difference, so as to distinguish the touch area, and the time features reveal the change trend of the touch area with time in the wet state, which provides a data basis for subsequent touch recognition in the wet mode, improves the accuracy of wet touch recognition and the effectiveness of data;

[0051] The present application learns the spatial features in depth through the wet mode touch recognition model, analyzes the spatial variation law of the spatial features, and learns the time features in depth through the wet mode touch recognition model, analyzes the time variation law of the time features, and finally recognizes the finger touch of the user in the wet mode based on the variation law through the wet mode touch recognition model, the advantage is that the spatial features and the time features in different situations can be learned through deep learning to determine the range of the spatial features and the time features, only when the spatial variation law and the time variation law are met at the same time, the correct recognition can be made, and the accuracy of wet touch recognition is greatly improved. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 It is a step flow chart of the method of the present application;

[0053] Figure 2 It is a capacitance distribution diagram of the present application;

[0054] Figure 3 It is a capacitance difference distribution diagram of the present application;

[0055] Figure 4 It is a schematic diagram of two spatial features of the present application;

[0056] Figure 5 It is a schematic diagram of overlapping SF2 and the combination of the present application;

[0057] Figure 6 It is a schematic diagram of the combination of the present application after updating after overlapping SF2. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] Example 1, please refer to Figure 1 As shown, this application provides a method for establishing a wet-pattern touch recognition model based on deep learning, including the following steps:

[0060] Step S1: Establish a wet-mode touch recognition model, configure test samples, and collect test data; Step S1 includes the following sub-steps:

[0061] Step S101: A wet mode touch recognition model is built using a convolutional neural network;

[0062] Step S102: Install a capacitive sensing matrix behind the touch screen to monitor the capacitive signal on the touch screen.

[0063] Step S103: Construct a humid environment group and a humid touch group. Pour water stains on the touch screen in the humid environment group and pour water stains on the touch screen in the humid touch group, and touch the touch screen with your finger.

[0064] Step S104: Number the capacitive sensors in the capacitive sensing matrix using the symbol CS. ij This represents the expression, where i and j are both positive integers and i and j are the indices of CS, and CS... ij This represents the capacitive sensor located in the i-th row and j-th column, arranged from top to bottom and from left to right.

[0065] Step S105, CS ij The detected capacitance signal is marked as CG. ij Simultaneously recording CG ij The monitoring time is numbered and marked as T in order from morning to night. n Where n is a positive integer and n is the index of T, the index of T is... n CG ij Let CC(n,i,j) be the index of CC, and CC(n,i,j) represent belonging to T. n CG ij ;

[0066] Step S106, record T for the humid environment group nand CC(n, i, j) are marked as T1 n and CC1(n, i, j) are named as wet environment data, recording T n and CC(n, i, j) are marked as T2 n and CC2(n, i, j) are named as wet touch data, the wet environment data and the wet touch data are test data.

[0067] In the embodiment, the capacitive sensing matrix is an existing capacitive sensing technology, which will not be described in detail in the embodiment, such as CS 12 13 , wherein i = 12, j = 13, representing the capacitive sensor at the 12th row and the 13th column, and the recorded capacitive signal is CG 12 13 , assuming that the capacitive signals within 0.5 seconds are collected, and the sampling frequency of the capacitive sensor is 1000Hz, that is, 1000 capacitive signals are collected within 1s, and 500 capacitive signals are collected within 0.5s, and the number is T n , 1≤n≤500, CG 12 13 in T1 is CC(1, 12, 13), and CC(1, 12, 13) in the wet environment group is CC1(1, 12, 13), and CC(1, 12, 13) in the wet touch group is CC2(1, 12, 13).

[0068] Step S2, feature collection is performed on the test data to obtain the spatial feature and the time feature of the finger touch under the wet environment; step S2 includes the following sub-steps:

[0069] Step S201, feature collection is performed on CC(n, i, j) in the test data to obtain the spatial feature of the finger touch under the wet environment;

[0070] Step S201 includes the following sub-steps:

[0071] Please refer to Figure 2 , step S201.1, CC1(n, i, j) and CC2(n, i, j) are obtained, and the same value of n is analyzed, a one-dimensional coordinate system is established with the capacitive signal as the horizontal axis, named as the capacitive distribution diagram, the capacitive signal is recorded into the capacitive distribution diagram, and the capacitive distribution diagram is analyzed to obtain the capacitive cluster.

[0072] Step S201.2, it is judged whether the number of the capacitive cluster is two, if yes, a clustering completion signal is output, if not, a clustering incomplete signal is output, if the clustering incomplete signal is output, the clustering coefficient of the clustering analysis is increased and the clustering analysis is performed again until the number of the capacitive cluster is two.

[0073] Step S201.3: Calculate the average value of the capacitance signal in the capacitance cluster and name it the average signal. Name the capacitance cluster with the largest average signal as the sensing cluster. Input CC1(n,i,j) and CC2(n,i,j) into the capacitance distribution map to analyze the sensing cluster. Name the sensing clusters corresponding to CC1(n,i,j) and CC2(n,i,j) as the sensing environment cluster and the sensing touch cluster, respectively.

[0074] Step S201.4: Calculate the average value of the capacitance signal in the sensed environment cluster, name it as the average environmental capacitance, and represent it by the symbol ACE. Mark the capacitance signal in the sensed touch cluster as STC(n,i,j), calculate STC(n,i,j)-ACE, and name the calculation result as capacitance difference.

[0075] In practice, each value of n is analyzed independently. Taking n=1 as an example, CC1(1,i,j) and CC2(1,i,j) are analyzed to construct the capacitance distribution diagram as follows. Figure 2 As shown, by Figure 2 It is clearly observable that the dense capacitance signals on the left constitute one capacitance cluster, and the continuous capacitance signals on the right constitute another capacitance cluster. At this time, there are two capacitance clusters. The clustering is completed, and the two capacitance clusters are output. The reason for dividing the screen into two clusters is that there are still some areas on the touch screen that are not in contact with liquid or fingers. The capacitance signals detected in these areas are usually 0 or very close to 0. The capacitance signals detected in areas with liquid and finger touch form a sensed cluster. CC1(n,i,j) and CC2(n,i,j) are analyzed separately to obtain the sensed environment cluster and the sensed touch cluster. The average environmental capacitance is calculated to be 30fF. The capacitance difference corresponding to each STC(1,i,j) is calculated. Due to the large amount of data, this embodiment does not show it in detail.

[0076] Step S201.5: Number the capacitance differences and label them as CD in ascending order of capacitance difference. m Where m is a positive integer and m is the index of CD, count the number of different capacitance differences, and divide CD... m The quantity is marked as CQ m ;

[0077] Please see Figure 3 As shown, in step S201.6, CD m For the X-axis, CQ m Establish a two-dimensional coordinate system for the Y-axis, named the capacitance difference distribution plot, and set CQ... m According to CD m Enter the capacitance difference distribution map and name the coordinate points in the capacitance difference distribution map as capacitance difference distribution points;

[0078] Step S201.7, connecting adjacent points of the distribution of the capacitance difference by a smooth curve, naming the finally connected curve as the distribution curve of the capacitance difference, obtaining the lowest point of the distribution curve of the capacitance difference as CD m , naming the capacitance boundary value;

[0079] Step S201.8, naming CC2(n, i, j) less than the capacitance boundary value as the capacitance ambient signal, naming CC2(n, i, j) greater than or equal to the capacitance boundary value as the capacitance touch signal, and naming the area where the capacitance touch signal corresponds as the capacitance touch area, that is, the spatial feature;

[0080] In a specific implementation, when CD m and CQ m are input, instead of inputting the data of a single test, the capacitance difference value when n = 1 in each of a large number of experimental data is analyzed to obtain a capacitance difference distribution graph, and the analysis process of each value of n is independently performed; the capacitance difference distribution graph is constructed as shown in Figure 3 , since the liquid also generates a capacitance signal on the touch screen, the finger touch in the liquid range will be affected, but the capacitance signal generated during the touch will still be slightly greater than the capacitance signal generated by the liquid, and there will be a boundary between them, that is, the capacitance boundary value in the capacitance difference distribution graph, the capacitance boundary value as the lowest point of the capacitance difference distribution curve, is usually the capacitance difference value monitored and calculated at the boundary of the finger touch, which is affected by the liquid and the finger touch, slightly greater than the capacitance signal generated by the liquid and slightly less than the capacitance signal generated by the finger in the liquid, and the number is very small, so the lowest point of the capacitance difference distribution curve is selected as the capacitance boundary value; the capacitance boundary value is obtained from Figure 3 , and CC2(1, i, j) greater than or equal to 36fF is named as the capacitance touch signal, the capacitance sensors exist in the touch screen in a matrix form, each capacitance sensor is pre-set with a rectangular area, and the area composed of the rectangular area where the capacitance sensor corresponding to the capacitance touch signal is located is the capacitance touch area, that is, the spatial feature;

[0081] Step S202, collecting the time feature of the finger touch in the humid environment from the test data as a whole;

[0082] Step S202 includes the following sub-steps:

[0083] Step S202.1, calculating the average value of the capacitance signal in the capacitance touch area, and naming it as the area average capacitance;

[0084] Step S202.2, marking the area average capacitance of T n as RAC n , and T nX-axis, RAC n A two-dimensional coordinate system is established for Y-axis, named as capacitance change graph, and RAC n According to T n The capacitance change graph is recorded;

[0085] Step S202.3, linear function fitting is performed on the capacitance change graph to obtain a capacitance change function, and a slope of the capacitance change function is obtained, named as capacitance change amplitude, and the capacitance change amplitude is the time characteristic;

[0086] In the specific implementation, since the data amount is too large, the regional average capacitance is not displayed in this embodiment, the capacitance change graph is constructed, and the capacitance change amplitude is obtained through linear function fitting, which is 0.0341. The capacitance change amplitude shows the change trend of the regional average capacitance with time. When the finger touches, as the finger approaches the touch screen, the regional average capacitance also increases, so there is a change trend of the regional average capacitance in the time dimension, that is, the capacitance change amplitude, and the capacitance change amplitude shows the time characteristic of the finger touching the screen in the wet area.

[0087] Step S3, deep learning is performed on the space characteristic and the time characteristic through the wet mode touch recognition model, and the change law of the space characteristic and the time characteristic is analyzed; step S3 includes the following sub-steps:

[0088] Step S301, deep learning is performed on the space characteristic through the wet mode touch recognition model, and the spatial change law of the space characteristic is analyzed;

[0089] Step S301 includes the following sub-steps:

[0090] Please refer to Figure 4 Fig. 1, step S301.1, all space characteristics in the wet touch data are collected, the space characteristics are numbered, and the symbol SF t is used to represent, wherein t is a positive integer and t is the serial number of SF;

[0091] Please refer to Figures 5 to 6 Fig. 1, step S301.2, SF1 is named as a combination, t=2 is started, SF t is overlapped with the combination, and it is required that when SF t is overlapped with the combination, the overlapping area of SF t and the combination reaches the maximum, the combination is updated as the contour of the overlapped image, t+1 is added and overlapped again until the maximum value of t is reached;

[0092] Step S301.3, the final combination is named as a space reference graph, the similarity of SF t and the space reference graph is calculated, and marked as SY t ;

[0093] Step S301.4, Statistical analysis of SY t The range of these is named the spatial variation law;

[0094] In specific implementation, for example, SF1 and SF2 are respectively Figure 4 The spatial features on the left and right sides are considered. In the initial state, the composite body is SF1, which represents the spatial features on the left. SF2 is then overlapped with the composite body, as shown below. Figure 5 As shown, when overlapping, it is necessary to ensure SF t Having the largest contact area with the composite, and then obtaining the updated composite after overlapping SF2, as shown below. Figure 6 As shown, by analogy and continuous overlap, a spatial reference map is finally obtained. The spatial reference map is a collection of spatial features from a large amount of experimental data. By analyzing the similarity between each spatial feature and the spatial reference map, the degree of similarity between the spatial feature and the spatial reference map can be determined. When a spatial feature that meets the conditions is detected, it can be confirmed that the touch area conforms to the spatial change law. The spatial change law is statistically obtained as [83.4%, 92.6%];

[0095] Step S302: Deep learning is performed on the time features through the wet mode touch recognition model to analyze the time variation pattern of the time features; specifically, the range of the time features is statistically analyzed, which is the time variation pattern.

[0096] In practice, the statistical results show that the time variation pattern is [0.018, 0.0364].

[0097] Step S4: Based on the changing patterns, the user's finger touch in the wet mode is recognized using a wet mode touch recognition model; Step S4 includes the following sub-steps:

[0098] Step S401: Monitor the capacitance signal on the touch screen in real time and analyze its spatial and temporal characteristics, which are named spatial real-time characteristics and temporal real-time characteristics, respectively.

[0099] Step S402: Calculate the similarity between the real-time spatial features and the spatial reference map, and name it the real-time spatial similarity. If the real-time spatial similarity is within the spatial change law, output the spatial verification pass signal; otherwise, output the spatial verification failure signal.

[0100] Step S403: Determine whether the real-time time feature is within the time change pattern. If so, output a time verification pass signal; otherwise, output a time verification failure signal.

[0101] Step S404: Determine whether the touch function is effective in the humid mode based on the output signal;

[0102] Step S404 includes the following sub-steps:

[0103] Step S404.1, if the space verification pass signal and the time verification pass signal are output at the same time, marking the touch in the wet mode as valid;

[0104] Step S404.2, if the space verification fail signal or the time verification fail signal is output, marking the touch in the wet mode as invalid;

[0105] In a specific implementation, the space real-time feature and the time real-time feature are updated in real time, and only when the space verification pass signal and the time verification pass signal are output at the same time, it is confirmed that the touch is a valid touch, rather than a false touch or a touch failure caused by liquid.

[0106] In an embodiment, the electronic device can include a processor, a communication interface, a memory, and a communication bus. The processor, the communication interface, and the memory can communicate with each other through the communication bus. The memory stores computer readable instructions. The processor can call the instructions in the memory. When the computer readable instructions are executed by the processor, the steps in the method for establishing a wet mode touch recognition model based on deep learning are executed to implement the following functions: establishing a wet mode touch recognition model, configuring test samples and collecting test data; collecting features of the test data, obtaining space features and time features of finger touch in a wet environment; deep learning of the space features and the time features through the wet mode touch recognition model, analyzing the change rule of the space features and the time features; and identifying finger touch of a user in a wet mode based on the change rule through the wet mode touch recognition model.

[0107] In addition, the logic instructions in the memory described above can be implemented in the form of a software functional unit and sold or used as an independent product. When used, the computer readable storage medium can store the logic instructions. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0108] In embodiment 3, the application further provides a computer program product, which comprises a computer program stored on a computer readable storage medium, and the computer program comprises program instructions, when the program instructions are executed by a computer, the computer can execute the deep learning based wet mode touch recognition model establishing method provided by each method, and the method comprises the following steps: establishing a wet mode touch recognition model, configuring test samples and collecting test data; collecting features of the test data, and acquiring spatial features and time features of finger touch in a wet environment; performing deep learning on the spatial features and the time features through the wet mode touch recognition model, analyzing change rules of the spatial features and the time features; and identifying finger touch of a user in a wet mode through the wet mode touch recognition model based on the change rules.

[0109] In embodiment 4, the application further provides a computer readable storage medium, and the application provides a storage medium, which stores a computer program, and the computer program is executed by a processor, and the steps in the deep learning based wet mode touch recognition model establishing method are executed to realize the following functions: establishing a wet mode touch recognition model, configuring test samples and collecting test data; collecting features of the test data, and acquiring spatial features and time features of finger touch in a wet environment; performing deep learning on the spatial features and the time features through the wet mode touch recognition model, analyzing change rules of the spatial features and the time features; and identifying finger touch of a user in a wet mode through the wet mode touch recognition model based on the change rules.

[0110] Through the description of the above embodiments, the embodiments of the application can be provided as a method, a system or a computer program product. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the method described in each embodiment or some parts of the embodiment.

[0111] In the embodiments provided by the application, it should be understood that the disclosed system or method can be implemented in other manners. The above described embodiments are merely exemplary, for example, the division of the modules or units is only a logical function division, and there can be another division manner in actual implementation, for example, a plurality of modules or units can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different parts can be indirect couplings or communication connections through some interfaces, and there can be electric, mechanical or other forms.

[0112] It should be noted that the above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the same. Although the present application has been described in detail with reference to the foregoing examples, it should be understood by those skilled in the art that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalent features. Such modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for establishing a wet mode touch recognition model based on deep learning, characterized in that, Includes the following steps: Establish a touch recognition model for a humid mode, configure test samples, and collect test data; Feature acquisition was performed on the test data to obtain the spatial and temporal features of finger touch in a humid environment; By using a wet-mode touch recognition model, deep learning is performed on spatial and temporal features to analyze the changing patterns of spatial and temporal features. Based on the changing patterns, a wet-mode touch recognition model is used to recognize users' finger touches in wet conditions; The process of establishing a touch recognition model for a humid environment, configuring test samples, and collecting test data includes the following sub-steps: A touch recognition model for wet conditions was built using a convolutional neural network. A capacitive sensing matrix is ​​installed behind the touchscreen to monitor the capacitive signals on the touchscreen. Construct a humid environment group and a humid touch group, pour water stains on the touch screen in the humid environment group, pour water stains on the touch screen in the humid touch group, and touch the touch screen with your finger; The capacitive sensors in the capacitive sensing matrix are numbered and represented by the symbol CSij, where i and j are both positive integers and i and j are the serial numbers of CS. CSij represents the capacitive sensor located in the i-th row and j-th column in order from top to bottom and from left to right. The capacitance signal monitored by CSij is marked as CGij, and the monitoring time of CGij is recorded. The monitoring time is numbered and marked as Tn in order from early to late, where n is a positive integer and n is the sequence number of T. CGij belonging to Tn is marked as CC(n,i,j), where (n,i,j) is the sequence number of CC, and CC(n,i,j) represents CGij belonging to Tn. Record Tn and CC(n,i,j) of the humid environment group, and label them as T1n and CC1(n,i,j) respectively. Name them as humid environment data. Record Tn and CC(n,i,j) of the humid touch group, and label them as T2n and CC2(n,i,j) respectively. Name them as humid touch data. The humid environment data and humid touch data are the test data. Feature acquisition of test data to obtain spatial and temporal features of finger touch in a humid environment includes the following sub-steps: Feature acquisition is performed on CC(n,i,j) in the test data to obtain the spatial features of finger touch in a humid environment; Feature collection was performed on the overall test data to obtain the temporal characteristics of finger touch in a humid environment; Feature acquisition of CC(n,i,j) in the test data to obtain spatial features of finger touch in a humid environment includes the following sub-steps: Obtain CC1(n,i,j) and CC2(n,i,j), and analyze the same value of n. Establish a one-dimensional coordinate system with the capacitance signal as the horizontal axis, and name it the capacitance distribution map. Input the capacitance signal into the capacitance distribution map, and perform cluster analysis on the capacitance distribution map to obtain the capacitance cluster. Determine if the number of capacitor clusters is two. If yes, output a clustering complete signal. If no, output a clustering incomplete signal. If an incomplete signal is output, increase the clustering coefficient and perform clustering analysis again until the number of capacitor clusters is two. Calculate the average value of the capacitance signal in the capacitance cluster and name it the average signal. Name the capacitance cluster with the largest average signal the sensing cluster. Input CC1(n,i,j) and CC2(n,i,j) into the capacitance distribution map to analyze the sensing cluster. Name the sensing clusters corresponding to CC1(n,i,j) and CC2(n,i,j) the sensing environment cluster and the sensing touch cluster, respectively. Calculate the average value of the capacitance signal in the sensed environment cluster, name it the average environmental capacitance, and represent it by the symbol ACE. Label the capacitance signal in the sensed touch cluster as STC(n,i,j), calculate STC(n,i,j)-ACE, and name the calculation result the capacitance difference. Number the capacitance differences and label them as CDm in ascending order, where m is a positive integer and m is the sequence number of CD. Count the number of different capacitance differences and label the number of CDm as CQm. Establish a two-dimensional coordinate system with CDm as the X-axis and CQm as the Y-axis, and name it the capacitance difference distribution map. Enter CQm into the capacitance difference distribution map according to CDm, and name the coordinate points in the capacitance difference distribution map the capacitance difference distribution points. Connect adjacent capacitance difference distribution points with a smooth curve, and name the resulting curve the capacitance difference distribution curve. Obtain the CDm corresponding to the lowest point of the capacitance difference distribution curve and name it the capacitance boundary value. CC2(n,i,j) less than the capacitance boundary value is named the capacitance environment signal, CC2(n,i,j) greater than or equal to the capacitance boundary value is named the capacitance touch signal, and the area where the capacitance sensor corresponding to the capacitance touch signal is located is named the capacitance touch area. The capacitance touch area is the spatial feature.

2. The method for establishing a wet pattern touch recognition model based on deep learning according to claim 1, characterized in that, The process of collecting features from the overall test data to obtain the temporal features of finger touch in a humid environment includes the following sub-steps: Calculate the average value of the capacitance signal in the capacitive touch area, and name it the area average capacitance. The average capacitance of region Tn is marked as RACn. A two-dimensional coordinate system is established with Tn as the X-axis and RACn as the Y-axis, named the capacitance change graph. RACn is entered into the capacitance change graph according to Tn. A linear function is fitted to the capacitance change graph to obtain the capacitance change function. The slope of the capacitance change function is obtained and named the capacitance change amplitude. The capacitance change amplitude is the time feature.

3. The method for establishing a wet pattern touch recognition model based on deep learning according to claim 2, characterized in that, The analysis of spatial and temporal feature variation patterns through deep learning using a humidity-pattern touch recognition model includes the following sub-steps: By using a wet-mode touch recognition model, we can perform deep learning on spatial features and analyze the spatial variation patterns of these features. By using a wet-mode touch recognition model to perform deep learning on time features, the temporal variation patterns of these features are analyzed.

4. The method for establishing a wet pattern touch recognition model based on deep learning according to claim 3, characterized in that, The analysis of spatial variation patterns of spatial features through deep learning using a humidity-pattern touch recognition model includes the following sub-steps: All spatial features in the wet touch data are collected and numbered using the symbol SFt, where t is a positive integer and t is the SF number; Name SF1 as the composite body. Starting at t=2, overlap SFt with the composite body. When overlapping, the overlap area between SFt and the composite body should reach the maximum. Update the composite body to the outline of the overlapping image. Add t+1 and overlap again until the maximum value of t is reached. The final composite is named the spatial reference graph, and the similarity between SFt and the spatial reference graph is calculated and labeled as SYt. The range of SYt is statistically analyzed and named the spatial variation law.

5. The method for establishing a wet pattern touch recognition model based on deep learning according to claim 4, characterized in that, By using a wet-mode touch recognition model to perform deep learning on time features, the time variation pattern of time features is analyzed, specifically the range of statistical time features, which is the time variation pattern.

6. The method for establishing a wet pattern touch recognition model based on deep learning according to claim 5, characterized in that, Based on the changing patterns, the wet-mode touch recognition model identifies user finger touches in wet conditions, including the following sub-steps: Real-time monitoring of the capacitive signal on the touch screen and analysis of its spatial and temporal characteristics are named spatial real-time characteristics and temporal real-time characteristics, respectively. Calculate the similarity between the real-time spatial features and the spatial reference map, and name it the real-time spatial similarity. If the real-time spatial similarity is within the spatial variation law, output a spatial verification pass signal; otherwise, output a spatial verification failure signal. Determine whether the real-time time feature is within the time change pattern. If so, output a time verification pass signal; otherwise, output a time verification failure signal. The effectiveness of touch control in humid mode is determined based on the output signal.

7. The method for establishing a wet pattern touch recognition model based on deep learning according to claim 6, characterized in that, Determining whether touch functionality is effective in humid conditions based on the output signal includes the following sub-steps: If both the space verification pass signal and the time verification pass signal are output simultaneously, then the touch in the wet mode is marked as valid; If the output space verification failure signal or time verification failure signal is output, then the touch in wet mode is marked as invalid.

Images