Gesture recognition method and system for power distribution automation terminal touch screen, electronic equipment and medium

Abstract

The invention relates to a gesture recognition method and system for a power distribution automation terminal touch screen, electronic equipment and a medium. The method comprises the steps that electric field distribution data provided by a deep electrode array are collected; gesture action features are extracted, amplitude-frequency joint correction is carried out based on a preset electric field intensity reference, and compressed delay time is obtained; sampling deep and shallow electrode arrays of the touch screen to obtain asynchronous acceleration sampling data; according to a deep electric field change parameter in the asynchronous acceleration sampling data, adjusting a deep electrode output voltage until the deep electric field change parameter converges, and then determining an identification delay compression value; and calculating gesture recognition optimization parameters according to the compression value and the converged electric field distribution data, carrying out time sequence alignment and voltage write-back on a plurality of target processing units, and finally outputting a gesture recognition result. According to the invention, low-delay and high-stability gesture recognition can be realized when the fingers of the user approach the touch screen, and the real-time performance and reliability of touch interaction are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent terminal control technology, and in particular to a gesture recognition method, system, electronic device and medium for a power distribution automation terminal touch screen. Background Technology

[0002] As the penetration rate of smart terminals in high-real-time scenarios such as industrial control, medical equipment, and automotive systems continues to rise, touch interaction has gradually replaced traditional mechanical buttons, becoming the "last mile" of human-machine command transmission. Gesture recognition is considered the "nerve ending" of touch interaction: its accuracy determines whether commands are correctly interpreted, and its response speed determines whether emergency operations can be executed promptly. Therefore, the "predictive gesture recognition" stage, which must be completed before the touchscreen surface even touches the device, has become a critical gate affecting the real-time performance and reliability of the entire system.

[0003] However, after long-term field testing and experimental analysis, researchers have found that existing predictive gesture recognition technologies still have significant blind spots when faced with weak electric field excitation: current mainstream solutions mostly rely on fixed threshold triggering or single-layer electrode sampling. These approaches essentially simplify "finger approach" to "single-point signal exceeding threshold," failing to fully consider the spatiotemporal diffusion characteristics of electric field disturbances during the finger approach phase, and also failing to provide stable and reproducible output results at the end of the recognition link. Specific limitations are reflected in: (1) When the operator’s finger approaches the touch screen but does not actually touch it, the electric field change amplitude captured by the shallow electrode array is only 3%–8% of the full scale. In addition, the effective signal is often submerged by random drift due to the coupling effects of ambient temperature and humidity, common-mode electromagnetic noise and stains on the panel surface. If the gain is directly amplified at this time, the common-mode disturbance will be amplified synchronously, resulting in the dilemma of “noise and signal increasing together”, which further increases the uncertainty of front-end resolution.

[0004] (2) The benchmark calibration data provided by the deep electrode, which serves as an important reference for assessing the credibility of the finger approach event, involves multi-point sampling, mean filtering, and compensation voltage iteration in its generation process, and naturally has a computational lag of hundreds of microseconds to several milliseconds. Meanwhile, the preprocessing process of the shallow electrode (impedance change rate detection, moving average denoising, and amplitude over-limit judgment) is often completed and triggered in the tens square (hundreds) microsecond range. The time "misalignment" between the two leads to the system falling into the predicament of "front end has been triggered, back end is not ready", forcing the main control unit to make a trade-off between "waiting for calibration" and "risky recognition", which ultimately manifests as the uncontrollable overall recognition delay.

[0005] (3) In high-concurrency scenarios, the touch screen needs to process multi-channel signals such as multi-finger approach, hover, and swipe gestures at the same time. If each channel waits for its own calibration to be completed before converging upwards, the parallel architecture will lose its parallel advantage due to "each doing its own thing". If the channel results are extrapolated and supplemented in the time domain, artifacts are easily introduced, causing the gesture trajectory to jump or miss segments, reducing the user experience. Summary of the Invention

[0006] In view of the above-mentioned deficiencies or disadvantages, the present invention provides a gesture recognition method, system, electronic device and medium for a power distribution automation terminal touch screen, which can solve at least one of the above technical problems.

[0007] In a first aspect, the present invention provides a gesture recognition method for a touch screen of a power distribution automation terminal, comprising: When the user's finger is within the preset detection distance range of the touch screen, the electric field distribution data on the surface of the touch screen is collected; the electric field distribution data includes the deep electric field variation parameters of the deep electrode array of the touch screen; Gesture motion features are extracted from electric field distribution data, and amplitude-frequency joint correction is performed on the gesture motion features based on a preset electric field strength benchmark. Asynchronous acceleration processing is performed on the sampling frequencies of the shallow and deep electrode arrays of the touch screen based on the compressed delay time, and asynchronous acceleration sampling data is obtained. The compressed delay time is obtained by performing joint amplitude and frequency correction on the gesture action features. The output voltage of the deep electrode array is adjusted according to the deep electric field change parameters in the asynchronous accelerated sampling data until the deep electric field change parameters converge, and the recognition delay compression value of the gesture action is determined from the current touch screen according to the electric field strength reference. The gesture recognition optimization parameters are calculated based on the recognition delay compression value and the current asynchronous accelerated sampling data, which includes the converged deep electric field change parameters. Based on the gesture recognition optimization parameters, clock alignment and voltage write-back operations are performed on multiple target processing units of the touch screen to output the gesture recognition results.

[0008] In a second aspect, the present invention provides a gesture recognition system for a touchscreen, comprising: The electric field distribution acquisition module is used to acquire electric field distribution data on the surface of the touch screen when the user's finger is within the preset detection distance range of the touch screen. The electric field distribution data includes the deep electric field change parameters of the deep electrode array of the touch screen. The gesture feature extraction module is used to extract gesture action features from electric field distribution data and perform amplitude-frequency joint correction on the gesture action features according to the preset electric field strength benchmark. The asynchronous sampling acceleration module is used to perform asynchronous acceleration processing on the sampling frequencies of the shallow electrode array and deep electrode array of the touch screen according to the compressed delay time, and to sample and obtain asynchronous accelerated sampling data. The compressed delay time is obtained by performing joint amplitude and frequency correction through gesture action features. The delay compression metering module is used to adjust the output voltage of the deep electrode array until the deep electric field change parameters converge based on the deep electric field change parameters in the asynchronous accelerated sampling data, and to determine the recognition delay compression value of the gesture action from the current touch screen based on the electric field strength reference. The gesture optimization measurement module is used to calculate gesture recognition optimization parameters based on the recognition delay compression value and the current asynchronous accelerated sampling data, which includes the converged deep electric field change parameters. The recognition result generation module is used to perform clock alignment and voltage write-back operations on multiple target processing units of the touch screen according to the gesture recognition optimization parameters, so as to output the gesture recognition result number.

[0009] Thirdly, the present invention provides an electronic device, comprising: At least one processor; and The memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor, which are executed by the at least one processor to enable the at least one processor to execute the gesture recognition method of any power distribution automation terminal touch screen of the present invention.

[0010] Fourthly, the present invention provides a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause a computer to execute the gesture recognition method of any power distribution automation terminal touch screen of the present invention.

[0011] The technical solution of this invention involves collecting deep electric field change parameters provided by a deep electrode array on the touchscreen surface when a user's finger approaches the touchscreen, forming electric field distribution data. Gesture features are then extracted from this data, and amplitude-frequency joint correction is performed on the features using a preset electric field strength benchmark to obtain a compressed delay time. Asynchronous acceleration processing is then performed on the sampling frequencies of the shallow and deep electrode arrays of the touchscreen based on this delay time, and asynchronous accelerated sampling data is obtained. The output voltage of the deep electrode array is adjusted according to the deep electric field change parameters in the asynchronous accelerated sampling data until the deep electric field change parameters converge, and the recognition delay compression value is determined using the electric field strength benchmark. Gesture recognition optimization parameters are then calculated based on the recognition delay compression value and the current asynchronous accelerated sampling data. Finally, clock alignment and voltage write-back operations are performed on multiple target processing units, and the gesture recognition result is output. This achieves low-latency, high-stability gesture recognition even when the finger is not in contact (i.e., when the user's finger is within a preset detection distance range of the touchscreen), improving the real-time performance and reliability of touch interaction. Attached Figure Description

[0012] Figure 1 This is a flowchart of a gesture recognition method for a power distribution automation terminal touch screen according to an embodiment of the present invention; Figure 2 This is a structural block diagram of a touchscreen gesture recognition system according to an embodiment of the present invention; Figure 3 This is a block diagram of an electronic device used to implement embodiments of the present invention. Detailed Implementation

[0013] The following description, in conjunction with the accompanying drawings, illustrates exemplary embodiments of the present invention, including various details to aid understanding. These details should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope of the invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0014] According to the first aspect, the present invention provides a gesture recognition method for a touch screen of a power distribution automation terminal. The method can be applied to a finger recognition system (hereinafter referred to as "system") of a power distribution automation terminal. The power distribution automation terminal can be a smartphone, tablet computer, vehicle central control unit, smartwatch, portable game console, VR (Virtual Reality) or AR (Augmented Reality) head-mounted display, interactive whiteboard, self-service ordering machine, access control tablet, and industrial handheld smart terminal device equipped with a touch screen. like Figure 1 As shown, the method may include: Step S110: When the user's finger is within the preset detection distance range of the touch screen, collect the electric field distribution data on the surface of the touch screen.

[0015] The electric field distribution data can include shallow electric field variation parameters of the shallow electrode array and deep electric field variation parameters of the deep electrode array of the touch screen. The shallow electric field variation parameters can include the capacitance change and impedance change rate between the electrodes, while the deep electric field variation parameters can include the compensation voltage value and the electric field intensity gradient.

[0016] Specifically, electric field distribution data can refer to a three-dimensional array consisting of the quantized electric field intensity values ​​output by all "shallow-deep electrode pairs" at a certain synchronous sampling time t, denoted as: ; in, The value of the second electric field intensity at the shallow electrode node (i,j) at time t (unit: millivolts); The electric field change (unit: millivolt) of the deep electrode node at the same coordinate after differential processing is required. The timing alignment accuracy requires a timestamp deviation of ≤0.5 milliseconds. The array size = number of rows M × number of columns N × 2 layers, with a uniform timestamp t for subsequent gesture feature extraction.

[0017] The system can use the differential sampling circuit configured in the power distribution automation terminal to input the signals of the shallow and deep electrodes to the ADC (Analog-to-Digital Converter) respectively, and add a timestamp to each sampling to ensure time consistency in subsequent processing.

[0018] Specifically, the shallow electrode array can be made of indium tin oxide (ITO) thin film and arranged 0.1 mm below the touch screen glass cover to sense minute capacitive disturbances when a finger approaches. The deep electrode array can be located 0.5 mm below the glass substrate and is mainly used to provide an electric field reference and interference compensation.

[0019] For example, the system can use a multi-channel capacitance sampling module integrated into the touchscreen control chip to simultaneously acquire the electric field intensity values ​​output by shallow and deep electrodes arranged in an 8×8 array at a sampling frequency of no less than 1000 Hz, forming an electric field distribution data frame. Specifically, before sampling, the system can first perform a 'no-field self-calibration' step, that is: continuously acquiring data for 200 milliseconds in a darkroom on the production line (temperature 23℃±2 degrees Celsius, relative humidity 45%~55%, electromagnetic background <0.3 mV), recording the average value for each deep node. with standard deviation ;Will Write it to the FLASH memory as the "zero field reference" for this node. This serves as the threshold for subsequent signal-to-noise assessment. This ensures the accuracy of subsequent differencing results. It truly reflects finger movement, rather than environmental drift.

[0020] Alternatively, during the initialization phase, the system can continuously collect electric field data for 100 milliseconds without the finger approaching, calculate the mean and standard deviation of each electrode node, use them as the initial values ​​of the electric field strength benchmark, and store them in non-volatile memory for subsequent calibration and comparison.

[0021] Step S120: Extract gesture features from electric field distribution data, and perform amplitude-frequency joint correction on gesture features according to preset electric field strength benchmark.

[0022] The gesture characteristics can include the amplitude of the electric field intensity change, the dominant frequency component, and the phase delay caused by the proximity of the fingers. A preset electric field intensity reference is also included. It can be generated in the following two-stage manner: In the static segment: air field measurements were taken concurrently with S110. As the zero point of amplitude; In the dynamic segment: whenever the system is idle (no trigger for 1 consecutive second), the running-average (moving average, which refers to continuously averaging the latest fixed-length sample in the data stream, and updating the result as new data arrives; used to smooth fluctuations and track slow changes) is updated at a rate of 64 frames / second. .

[0023] In actual calibration, amplitude correction can be performed using... To ensure consistency; frequency correction can be performed using approximately 1 second. Normalization is performed to account for slow environmental changes. This balances the needs of both absolute benchmark and tracking drift.

[0024] For example, the system can perform a Fast Fourier Transform (FFT) on the electric field distribution data to extract the main frequency components in the frequency band from 1 Hz to 30 Hz, and obtain the peak value of its envelope as the action amplitude through Hilbert transform. At the same time, it can calculate the cross-correlation function between adjacent electrode channels and extract the phase delay time.

[0025] For example, if a channel measures an electric field intensity variation of 120 mV, a dominant frequency of 12 Hz, and a phase delay of 3 ms, the system compares the electric field distribution data of this channel with a preset electric field intensity benchmark (e.g., an amplitude benchmark of 100 mV and a frequency benchmark of 10 Hz), and calculates correction coefficients: amplitude correction coefficient = 100 / 120 = 0.83, frequency correction coefficient = 10 / 12 = 0.83. Then, the measured features are linearly corrected to obtain standardized gesture features for subsequent recognition processing. Referring to the preset amplitude benchmark (100 mV) and frequency benchmark (10 Hz), the system performs joint parameterized correction on the original measured values ​​(amplitude 120 mV, dominant frequency 12 Hz) and phase delay (3 ms) of this channel. The correction process first calculates the amplitude correction coefficient and the frequency correction coefficient separately, and then, based on this coefficient cluster and phase delay information, generates a normalized gesture feature vector through feature space mapping for subsequent recognition module processing. The aforementioned channel refers to a signal transmission path that is independently sampled and processed in the touch screen electric field distribution acquisition system. This path corresponds to an "electrode pair" and its associated analog front-end circuit, which is composed of a specific shallow electrode node and a specific deep electrode node. Logically, it is assigned a unique channel number to carry the electric field change information at that spatial location.

[0026] Step S130: Perform asynchronous acceleration processing on the sampling frequencies of the shallow electrode array and deep electrode array of the touch screen according to the compressed delay time, and obtain asynchronous acceleration sampling data.

[0027] The compressed delay time is obtained by jointly correcting the gesture action features based on the amplitude and frequency of the execution.

[0028] Specifically, the sampling frequency can refer to the number of analog-to-digital conversions performed on a continuous electric field signal per unit time, denoted as . The unit is "Hertz (Hz)". The system's default value can be 1000 Hz, meaning a sample is completed every 1 millisecond; after asynchronous acceleration processing, the shallow electrode array... The frequency was increased to 2000 Hz, while the deep electrode array remained at 1000 Hz. The compressed delay time can refer to the absolute reduction in the end-to-end recognition delay measured by the system relative to the original delay after the gesture motion features have sequentially completed "amplitude-frequency joint correction," "asynchronous accelerated sampling," and "deep voltage convergence." This reduction is denoted as... The unit is "millisecond".

[0029] Specifically, the compressed delay time ( This refers to the difference between the total latency of the unaccelerated link and the total latency of the accelerated link, expressed in milliseconds; the calculation formula is as follows: ; in, The total delay before correction, This represents the total latency after acceleration and convergence processing; this value is used as a threshold to determine whether asynchronous acceleration and subsequent optimization parameter generation are triggered.

[0030] The system can evaluate the response delay of the corrected gesture features. If the delay time is lower than a preset threshold (such as 10 milliseconds), the sampling frequency of the shallow electrode array is increased to 2000 Hz, while the deep electrode array is kept at 1000 Hz, forming an asynchronous sampling strategy to balance processing efficiency and resource consumption.

[0031] For example, in the context of in-vehicle central control, when a user's finger quickly swipes across the touchscreen, the system detects that the corrected delay time is 8 milliseconds and triggers an asynchronous acceleration mechanism. The shallow electrodes collect data at a frequency of 2000 Hz, and the deep electrodes collect data at a frequency of 1000 Hz, thereby improving the front-end response speed of gesture recognition and avoiding deep data redundancy.

[0032] Step S140: Adjust the output voltage of the deep electrode array according to the deep electric field change parameters in the asynchronous accelerated sampling data until the deep electric field change parameters converge, and determine the recognition delay compression value of the gesture action from the current touch screen according to the electric field strength reference.

[0033] The deep electric field change parameters refer to the quantized set collected by each node of the deep electrode array and used to characterize the electric field compensation state during the process of a finger approaching or gliding across the touchscreen, including: compensation voltage value. Unit: millivolt; voltage difference between adjacent nodes Unit: millivolt; electric field intensity gradient The unit is millivolts per millimeter; the aforementioned three types of parameters can be indexed by channel number k, corresponding one-to-one with the shallow electric field change parameters, together constituting the "deep" dimension in the electric field distribution data.

[0034] Next, the convergence determination can be performed using the following double threshold and continuous counting mechanism: Voltage threshold: millivolt; Gradient threshold: Voltage difference between adjacent nodes The maximum fluctuation is ≤10 millivolts; The system can be confirmed to have converged only if it meets the voltage threshold and gradient threshold conditions mentioned above for three consecutive sampling periods (1 millisecond per period).

[0035] Once the condition is met, the output voltage of that node is locked, and the "converged" flag is set (using 1 bit to indicate that the electric field compensation of that electrode node or channel has met the standard); the system only enters the state when all valid node flags are 1. Calculations are performed to prevent premature quantization delays.

[0036] Then, identify the delayed compression value ( () refers to the relative proportion of delayed compression, with a value ranging from 0 to 1, denoted as Dimensionless It is a relative measure of the degree to which gesture recognition latency is compressed. The formula for calculating this recognition latency compression value is as follows:

[0037] in, The end-to-end identification delay (in milliseconds) before asynchronous acceleration and voltage convergence were employed. The actual measured delay after adoption; The value ranges from 0 to 1. A larger value indicates a more significant delay compression effect and is used to generate subsequent gesture recognition optimization parameters.

[0038] Specifically, the system can monitor the compensation voltage deviation in the deep electric field change parameters in real time through a closed-loop feedback control mechanism. If the deviation exceeds ±20 millivolts, the deep electrode output voltage is adjusted in steps of 0.1 volts until the electric field strength returns to the reference range, which is then determined to be a convergence state.

[0039] Alternatively, the system can record the time interval from the start of adjustment to the completion of convergence as the compensation response time, and combine it with the original recognition delay to calculate the recognition delay compression value = (original delay - compressed delay) / original delay. For example, when the compensation response time is 25 milliseconds, the original recognition delay is 100 milliseconds, and the compressed delay is 60 milliseconds, then the recognition delay compression value is (100-60) / 100 = 0.4, indicating a delay compression rate of 40%.

[0040] Step S150: Calculate the gesture recognition optimization parameters based on the recognition delay compression value and the current asynchronous accelerated sampling data.

[0041] The current asynchronous accelerated sampling data can include converged deep electric field variation parameters and shallow electric field variation parameters of the shallow electrode array obtained through asynchronous accelerated sampling. Gesture recognition optimization parameters can include weighting coefficients, filter bandwidth, and feature extraction window length for subsequent processing. The converged electric field distribution data refers to the instantaneous full-field electric field state set recorded after the system determines that the deep electric field variation parameters of all effective channels have converged (see the aforementioned definition of "convergence of deep electric field variation parameters"), including: the shallow electric field intensity of each channel. (Unit: millivolts); Deep compensation voltage for each channel (Unit: millivolts); Signal-to-noise ratio (SNR) (kΩ) for each channel (unit: decibels); Uniform timestamp (Unit: milliseconds). This dataset is considered a "stable frame" and is used for subsequent calculations of recognition latency compression values ​​and gesture recognition optimization parameters.

[0042] Next, the gesture recognition optimization parameters can refer to the converged electric field distribution data and the recognition delay compression value. A set of quantized control quantities obtained through joint calculation, used for dynamically configuring subsequent identification processes.

[0043] Specifically, the system can normalize the identification delay compression value to the interval of 0 to 1, and perform weighted fusion with the signal-to-noise ratio in the converged electric field distribution data to calculate the optimization parameters: The weighting coefficient is set to 0.5 × normalized compression value + 0.5 × signal-to-noise ratio coefficient. The filter bandwidth is set to 12 Hz to 25 Hz, and the feature extraction window length is set to 256 sampling points.

[0044] Specifically, the weighting coefficient formula was obtained through orthogonal fitting with 50 subjects and three humidity gradients (30% / 50% / 70%RH). Using the "recognition error rate" as the response value, DoE (Design of Experiments) analysis showed that the compression value and SNR (signal-to-noise ratio) contributed 48% and 46% to the error rate, respectively, hence equal weighting was applied; the remaining 6% was an interaction term and could be ignored. Therefore, a weighting of 0.5 / 0.5 was used, simplifying the calculation while maintaining an error rate increase of ≤3%.

[0045] RH (Relative Humidity) represents the percentage of water vapor content in the air compared to the saturated water vapor content. DoE analysis refers to the process of systematically changing and testing multiple variables to determine their impact on the results, and is often used to optimize process parameters.

[0046] For example, if the normalized compression value is 0.4 and the signal-to-noise ratio coefficient is 0.8, then the weighting coefficient = 0.5 × 0.4 + 0.5 × 0.8 = 0.6, which is used for weighting processing during subsequent multi-channel signal fusion.

[0047] Step S160: Perform clock alignment and voltage write-back operations on multiple target processing units of the touch screen according to the gesture recognition optimization parameters to output the gesture recognition result.

[0048] The target processing unit refers to the four smallest functional entities in the touchscreen gesture recognition link that are managed by an independent clock domain or power domain and can be dynamically reconfigured according to gesture recognition optimization parameters. Each target processing unit may include a signal preprocessing unit, a feature extraction unit, a recognition decision unit, and an output interface unit. The relationship between the target processing unit and the electrode node is "many-to-one mapping, hierarchical processing, and no direct physical connection".

[0049] Specifically, the system can use the weight coefficients in the gesture recognition optimization parameters as a benchmark to adjust the data buffer capacity and processing timing of each processing unit, so that the output time of all units is aligned with a unified clock reference, and write the optimized compensation voltage value back to the deep electrode array to ensure the stability of the electric field.

[0050] Specifically, the system can perform clock alignment and voltage write-back operations through the following steps 1 and 2: Step 1: The system first analyzes the clock skew. The FIFO (First In, First Out) depth is adjusted step by step in the order of downstream to upstream to align the output delay of the signal preprocessing unit, feature extraction unit, recognition decision unit, and output interface unit with the 2000 Hz master clock edge, with a residual error ≤ 1 millisecond.

[0051] Step 2: After the clock domain is locked, then use the locked time... As the initial error, substitute the dynamic voltage regulation coefficient. Perform the final closed-loop iteration; at this point, the clocks are aligned, ensuring that all nodes refresh their voltages simultaneously, thus eliminating transient electric field artifacts caused by scattered refresh times.

[0052] For example, the system recognizes a gesture as "clockwise rotation", with a trajectory coordinate sequence of [(160,180),(170,170),(180,160)], an action type identifier of "rotation", a response time of 55 milliseconds, and finally outputs a structured recognition result: {"Gesture type": "clockwise rotation", "Trajectory coordinates": see above, "Response time": 55 milliseconds, "Confidence": 0.92}, which can be called by upper-layer applications.

[0053] Therefore, according to the above implementation, the system can collect the deep electric field change parameters provided by the deep electrode array on the touch screen surface when the user's finger approaches the touch screen, forming electric field distribution data; then extract gesture action features from this data, and perform amplitude and frequency joint correction on the features using a preset electric field strength benchmark to obtain a compressed delay time; then perform asynchronous acceleration processing on the sampling frequency of the shallow electrode array and the deep electrode array of the touch screen based on this delay time, and sample to obtain asynchronous acceleration sampling data. The output voltage of the deep electrode array is adjusted according to the deep electric field change parameters in the asynchronous acceleration sampling data until the deep electric field change parameters converge, and the recognition delay compression value is determined using the electric field strength benchmark; then, gesture recognition optimization parameters are calculated based on the recognition delay compression value and the current asynchronous acceleration sampling data; finally, clock alignment and voltage write-back operations are performed on multiple target processing units, and the gesture recognition result is output. Thus, low-latency, high-stability gesture recognition can be achieved even before the finger touches the screen, improving the real-time performance and reliability of touch interaction.

[0054] Specifically, by using the methods described above, the three major shortcomings of traditional solutions—"weak signals being overwhelmed by noise, time misalignment between shallow and deep layers, and chaotic high-concurrency channels"—can be resolved before a finger touches the touchscreen. The specific comparative effects are as follows: 1. Addressing the drawback of weak electric field signals being easily submerged by noise: Existing technologies generally have a signal-to-noise ratio of less than 3 dB and a false recognition rate of more than 10%; The above embodiment improves the effective signal-to-noise ratio to ≥6 dB through two-stage processing of "amplitude and frequency joint correction + signal-to-noise ratio weighted filtering", and the measured false recognition rate is reduced by 40%, significantly reducing false triggering caused by environmental interference.

[0055] 2. Addressing the time misalignment defect of "front-end triggering, back-end not ready" between shallow and deep electrodes: Existing technologies often experience a recognition delay of 100 milliseconds due to the same sampling frequency between shallow and deep layers; The above embodiment adopts "asynchronous accelerated sampling" (2000 Hz for shallow layers / 1000 Hz for deep layers), and combines it with FIFO resampling and clock alignment, compressing the recognition delay to ≤50 milliseconds, eliminating the trajectory lag caused by the asynchrony of shallow and deep layer data.

[0056] 3. Addressing the shortcomings of high-concurrency multi-channel output timing deviation: Existing technologies can cause output timing discrepancies of more than 5 milliseconds when multiple channels are in parallel, resulting in abrupt changes in gesture trajectory; This invention achieves clock alignment through "FIFO buffer + dynamic adjustment of buffer capacity", controlling the timing deviation of 64-channel output to ≤1 millisecond, effectively avoiding trajectory breakage or coordinate jumps, and ensuring continuous and smooth sliding trajectory.

[0057] In fact, the above embodiments have the beneficial effects shown in Table 1 below, which address the problems of weak signals being submerged by noise, time misalignment between deep and shallow layers, and chaotic high-concurrency channels in existing technologies.

[0058] Table 1. Beneficial Effects Data Table

[0059] In some embodiments, collecting electric field distribution data on the surface of the touch screen includes: Multiple first electric field intensity values ​​are obtained from each electrode node of the shallow electrode array, and a shallow reference electric field is constructed based on each first electric field intensity value.

[0060] The first electric field strength value can be obtained by taking the arithmetic mean of electric field strength data collected continuously for 10 seconds at a sampling rate of 1000 Hz during the system initialization phase and under static conditions where the user's finger is not near the touchscreen. The unit is millivolts. The shallow reference electric field can be a two-dimensional matrix composed of all the first electric field strength values. This serves as the zero reference base for subsequent difference operations.

[0061] For example, after the power distribution automation terminal is powered on, in a static state where no one touches the screen, the main control chip of the power distribution automation terminal sequentially selects 8 rows and 8 columns of shallow electrode nodes; each node continuously collects electric field strength 10 times at a frequency of 1000 Hz, resulting in a total of 640 raw data points; then, the maximum and minimum values ​​of each group are removed, and the arithmetic mean of the remaining data is taken to calculate 64 average electric field strength values; finally, the aforementioned average values ​​are written into the first 64 addresses of flash memory in row and column order, thus forming a shallow reference electric field matrix, which can still be retained after power failure and can be directly called for subsequent differential operations.

[0062] The impedance change rate of the shallow electrode array is obtained in response to the user's finger being within the preset detection distance range of the touch screen.

[0063] The impedance change rate can refer to the difference between the electrode node impedance measured in the current sampling period and the reference impedance, divided by the sampling interval time, with the unit being ohms per millisecond.

[0064] For example, the system calculates the impedance change rate of each node in real time with a sliding window of 2 milliseconds. If the impedance change rate of any node is greater than the preset trigger condition value of 1000 ohms per millisecond, it is determined to be a valid proximity event.

[0065] If the impedance change rate exceeds the preset trigger condition value, multiple second electric field intensity values ​​are obtained from each electrode node of the shallow electrode array, and multiple third electric field intensity values ​​are obtained simultaneously from each electrode node of the deep electrode array.

[0066] The second electric field strength value can be: a real-time electric field strength sequence continuously acquired by the system from the shallow electrode node at a sampling rate of 2000 Hz after the triggering condition is met, in millivolts. The third electric field strength value can be: an electric field strength sequence synchronously acquired by the system from the deep electrode node at the corresponding row and column coordinates at the same timestamp, at a sampling rate of 1000 Hz, in millivolts.

[0067] Differential processing is performed on each third electric field intensity value based on the shallow reference electric field to generate deep electric field variation parameters.

[0068] When performing differential processing, the system calculates the third electric field strength value for each channel k. With the corresponding shallow reference electric field The difference = and will With corresponding deep compensation voltage These parameters together constitute the deep electric field variation parameters, measured in millivolts. Used to characterize the amount of compensation required for deep electrodes to withstand environmental disturbances.

[0069] The electric field distribution data is obtained by performing time-series alignment between each second electric field intensity value and each third electric field intensity value after differential processing.

[0070] When performing timing alignment, the system can use the shallow sampling clock as a reference and resample the deep electric field variation parameters to 2000 Hz through an asynchronous FIFO buffer, so that each second electric field intensity value The third electric field strength value after the corresponding difference Data pairs are formed at the same timestamp t, and finally an electric field distribution data frame containing all channels and double-layer electric field information is constructed for subsequent gesture feature extraction.

[0071] Therefore, according to the above implementation, the system can trigger synchronous sampling by shallow impedance change rate when the finger is not touching the screen, use shallow reference electric field to differentially drift the deep signal, and form highly stable, low-latency electric field distribution data through time alignment, providing a reliable input basis for subsequent gesture recognition.

[0072] In some embodiments, hand gesture features are extracted from electric field distribution data, including: A frequency domain transformation is performed on the electric field distribution data to obtain the signal-to-noise ratio values ​​for each electrode node.

[0073] When performing frequency domain transformation, the system can process the time-series electric field signal of each channel k. Perform a 256-point Fast Fourier Transform and take the signal power in the frequency band from 1 Hz to 30 Hz. With noise power and The signal-to-noise ratio value was calculated. The unit is decibel; if SNR(k) ≥ preset signal-to-noise threshold of 6 dB, then the channel is marked as "enhanced electric field signal channel".

[0074] Electrode nodes with a signal-to-noise ratio higher than a preset signal-to-noise threshold are used as enhanced electric field signal channels. The signal components of each enhanced electric field signal channel are filtered, and an enhanced electric field signal set is constructed based on the filtered enhanced electric field signal channels.

[0075] When performing filtering, the system can use a fourth-order Butterworth bandpass filter to retain the 5Hz to 25Hz components and suppress high-frequency noise and baseline drift; the filtered signal is denoted as... All enhanced channels Constitutes a set of enhanced electric field signals .

[0076] The weighting coefficients are calculated based on the ratio of each signal-to-noise ratio value to a preset reference signal-to-noise ratio value. Then, the corresponding enhanced electric field signal channels in the enhanced electric field signal set are weighted according to each weighting coefficient to obtain multiple weighted electric field signals.

[0077] The preset baseline signal-to-noise ratio can be set to 6 dB; the system can be calculated using... The weighting coefficients are calculated; if If so, it is truncated to 1.5.

[0078] Next, the system can use calculation formulas The enhanced electric field signal channels in the enhanced electric field signal set are weighted for subsequent feature extraction; among them, It is the time-domain signal strength of the k-th enhanced electric field signal channel after bandpass filtering. It is the time-domain signal strength of the k-th enhancement channel after signal-to-noise ratio weighting.

[0079] Normalization is performed on each weighted electric field signal to generate a set of electric field signal components.

[0080] During normalization, the system can perform normalization on each... The sequence is scaled proportionally to the maximum value, so that || ||∞=1, thus obtaining the normalized electric field signal. Then all Arranged in order of channel number, forming a set of electric field signal components. , where K is the total number of enhancement channels.

[0081] Envelope features, phase features, and periodic features are extracted from the set of electric field signal components.

[0082] Among them, envelope features can refer to... Perform a Hilbert transform, take the analytic signal magnitude, and obtain the instantaneous envelope. Phase characteristics can be used to calculate the peak position of the cross-correlation function between adjacent channels, thus obtaining the phase delay between channels. The unit is milliseconds; the periodicity characteristic can be obtained by analyzing... Perform autocorrelation calculation and take the position of the first non-zero peak. The result is in milliseconds.

[0083] Specifically, envelope feature extraction may include the following steps: For the set of electric field signal components Normalized signal for each channel Performing the Hilbert transform yields the following analytic signal: H(k,t)= +j·Hilbert{}; Modulo value , is defined as the instantaneous envelope of the channel; All channels Calculate the arithmetic mean to obtain the mean envelope. The maximum value of is used as the envelope feature A, with the unit being millivolts; where mean is a function for calculating the arithmetic mean, used to calculate the sum of a set of values ​​divided by the number of values.

[0084] Phase feature extraction may include the following steps: Select spatially adjacent enhancement channels k and k+1, and calculate the cross-correlation function using the following formula: ; extract peak position , defined as inter-channel phase delay; For all available Calculate the arithmetic mean to obtain the phase characteristics. The unit is milliseconds.

[0085] Periodic feature extraction may include the following steps: For the average envelope Perform autocorrelation calculation:

[0086] Take the position of the first non-zero peak. , is defined as the gesture cycle; by Calculate the periodic characteristics, in Hertz.

[0087] The maximum amplitude of the envelope feature is taken as the gesture amplitude, the time offset of the phase feature is taken as the gesture phase, and the reciprocal of the period feature is taken as the gesture frequency to obtain the gesture features.

[0088] The system can set the amplitude of the gesture to a value. The unit is millivolts (mV). `max` is a function that takes the maximum value from a set of values; the gesture phase can take values ​​of... The unit is milliseconds; while the frequency of gesture actions can take values ​​of... The unit is Hertz; the three elements constitute the gesture action feature vector. This is for use by subsequent recognition models.

[0089] Therefore, according to the above implementation method, the system can quickly obtain stable and quantified gesture features in the stage before the finger touches the object, by using frequency domain signal-to-noise ratio filtering, weighting and normalized envelope-phase-period extraction, laying a data foundation for low-latency and high-robust gesture recognition.

[0090] In some embodiments, amplitude-frequency joint correction is performed on the gesture characteristics based on a preset electric field strength reference, including: Obtain a preset electric field strength reference, which includes the amplitude and frequency of a standard hand gesture.

[0091] The standard gesture range can be denoted as: Defined as: the maximum value measured by the 64-channel average envelope when a standard finger model slides in a straight line across the center area of ​​the touchscreen at a speed of 250 millimeters per second, in millivolts, written to flash memory before leaving the factory; while the standard gesture frequency can be denoted as... Defined as: the reciprocal of the period obtained from the autocorrelation peak of the average envelope under the same standard sliding action, in Hertz, with a typical value of 8 Hertz, also stored in flash memory for subsequent ratio correction.

[0092] Specifically, The calibration environment can be: temperature 23℃±2 degrees Celsius, relative humidity 45%~55%, electromagnetic background noise <0.3 mV; The calibration tool can be: a standard analog finger (diameter 18±1 mm, conductivity 5 S / m, i.e. 5 Siemens per meter), with a movement speed of 250±10 mm / s; The calibration steps can be as follows: In a darkroom environment, control a standard physical finger model to slide along the center of the touchscreen 10 times, and collect the maximum value of the 64-channel average envelope as... (Typical value 100 mV), the reciprocal of the peak period of the average envelope autocorrelation is used as... (Typical value 8 Hz), and will , Write to non-volatile memory; The calibration cycle can be: after every 1000 gesture recognitions, an empty field calibration is automatically performed (continuously collected for 200 milliseconds, updating the baseline value by ±5%).

[0093] The first ratio of the gesture amplitude to the standard gesture amplitude is calculated. The gesture amplitude is then multiplied by the first ratio to obtain the corrected amplitude.

[0094] The first ratio can be ;like If the value is greater than 1.5, then the amplitude limit is 1.5; the system's amplitude after correction is... The unit is millivolt; this multiplication operation can normalize the measured amplitude to a standard scale and eliminate the drift caused by differences in finger size or clothing.

[0095] The second ratio of the gesture frequency to the standard gesture frequency is calculated. The gesture frequency is then multiplied by the second ratio to obtain the corrected gesture frequency.

[0096] The second ratio can be ;like If the value is greater than 2, then the limit is 2; the system's operating frequency after correction is... The unit is Hertz; this step maps the frequency increase caused by rapid sliding to a standard range, ensuring the consistency of subsequent model inputs.

[0097] The corrected amplitude and frequency of the movement are used as the characteristics of the corrected gesture to complete the joint amplitude and frequency correction.

[0098] The expression for the corrected gesture motion feature vector output by the system can be: This is used for subsequent compression delay and identification processes.

[0099] Assuming the measured value is A = 120 millivolts and f = 10 Hz, =100 millivolts =8 Hz, then the system can calculate: =120 × 1.2 = 144 millivolts Hertz, forming a unified correction result.

[0100] Therefore, according to the above implementation method, the system can use the factory-calibrated standard gesture amplitude and frequency reference to complete the amplitude and frequency joint correction through a simple ratio multiplication stage before the finger touches the screen, thereby eliminating individual differences and environmental drift, and providing consistent and reliable feature input for subsequent low-latency recognition.

[0101] In some embodiments, asynchronous acceleration processing is performed on the sampling frequencies of the shallow electrode array and the deep electrode array of the touch screen according to the compressed delay time, and asynchronous acceleration sampling data is obtained, including: The delay level of each electrode node in the shallow and deep electrode arrays is determined based on the compressed delay time.

[0102] The latency level can be: based on the compressed latency time. The independent variable is the discrete level L quantized with a step size of 5 milliseconds. The calculation formula is... The smaller the L value, the shorter the processing latency of the channel, and the faster it should be. Therefore, the system can calculate L for each of the M×N electrode nodes and write it into the node configuration register as the basis for subsequent graded sampling; where floor is a function that rounds down, used to discard the decimal part of a real number and return the largest integer not greater than that number.

[0103] Specifically, the compressed delay time The following formula can be used for calculation:

[0104]

[0105] ; in, This refers to the time required for the entire identification process in the original (unaccelerated) solution. This refers to the time required for the entire process after employing asynchronous acceleration and parallel processing. The required time for shallow processing is fixed at 1 millisecond; 3 milliseconds are required for deep processing; This is equivalent to the 2 milliseconds required for traditional serial processing. To meet the processing requirements after asynchronous parallelism, the time has been reduced to 1 millisecond.

[0106] Substituting into the above formula, we get: .

[0107] like If the sampling time is ≥1 millisecond, subsequent asynchronous acceleration is allowed; otherwise, constant sampling is maintained to ensure the signal-to-noise ratio.

[0108] Electrode nodes in the shallow electrode array with a delay level lower than a preset threshold are designated as high-speed sampling channels, while electrode nodes in the deep electrode array with a delay level not lower than a preset delay threshold are designated as constant-speed sampling channels.

[0109] Specifically, the system can set the preset level threshold value to... If the L of a shallow node SL(i,j) is less than 3, it is marked as a "high-speed sampling channel"; otherwise, it is marked as a "normal-speed sampling channel". Similarly, if the L of a deep node DL(i,j) is greater than or equal to 3, it is retained as a "normal-speed sampling channel"; otherwise, it is downgraded to a "low-speed sampling channel". This hierarchical strategy ensures that the shallow fast-response region obtains higher temporal resolution, while the deep compensation region maintains a moderate sampling rate to save power consumption.

[0110] Increase the sampling frequency of the high-speed sampling channel, maintain the sampling frequency of the constant-speed sampling channel, and reduce the sampling frequency of the remaining electrode nodes.

[0111] Specifically, the system can increase the frequency of the high-speed sampling channel from 1000 Hz to 2000 Hz; keep the constant-speed sampling channel at 1000 Hz; and reduce the frequency of the remaining low-speed channels to 500 Hz. Therefore, the three frequencies are generated by independent clock dividers, forming an asynchronous sampling relationship of "shallow fast - deep stable - redundant slow". Cross-clock domain data alignment is achieved using an asynchronous FIFO buffer within the FPGA (Field-Programmable Gate Array) to avoid data loss.

[0112] The shallow electrode array and the deep electrode array are restarted, and asynchronous accelerated sampling data are acquired by the high-speed sampling channel and the constant-speed sampling channel in parallel output in the same clock domain.

[0113] When the shallow and deep electrode arrays restart, the main control chip of the power distribution automation terminal first issues a hierarchical configuration table, and then simultaneously starts three sets of clocks. Next, the system writes high-speed channel data into a FIFO via a 2000 Hz clock, and normal-speed channel data into another FIFO via a 1000 Hz clock. Finally, the data is retrieved by a unified 2000 Hz readout clock, forming asynchronous accelerated sampling data frames that are output in parallel within the same clock domain. This is used for subsequent electric field convergence and delay compression calculations.

[0114] Therefore, according to the above implementation method, the system can classify the electrode nodes into delay levels based on the compressed delay time during the non-finger contact stage, and achieve shallow high-speed and deep stable parallel sampling by asynchronous frequency boosting and FIFO cross-clock domain alignment, thereby reducing the overall recognition delay and saving computing resources.

[0115] In some embodiments, adjusting the output voltage of the deep electrode array to convergence of the deep electric field change parameters based on the deep electric field change parameters in the asynchronous accelerated sampling data, and determining the recognition delay compression value of the gesture action from the current touch screen based on the electric field strength reference, includes: The compensation response time corresponding to the deep electric field change parameters of each electrode node in the deep electrode array is calculated based on the asynchronous accelerated sampling data.

[0116] The compensation response time can be denoted as: Defined as: from the time the deep electrode node DL(k) receives the compensation voltage command to the node The time taken for three consecutive sample values ​​to fall within the ±20 mV stable range, in milliseconds.

[0117] For example, the system can monitor at a rate of 2000 Hz. And record using a hardware timer With an accuracy of 1 millisecond.

[0118] The voltage regulation step size of each electrode node in the deep electrode array is adjusted based on the comparison results between each compensation response time and the preset response time interval, thereby generating dynamic voltage regulation coefficients for multiple electrode nodes.

[0119] Among them, the preset response time range can be taken as follows: milliseconds; if If the time is less than 20 milliseconds, the system determines that the node is over-adjusted, and the voltage adjustment step size is adjusted accordingly. Reduce from 0.1 volts to 0.05 volts; if If the time interval is greater than 50 milliseconds, the system will determine that the node is under-adjusted. Increase to 0.3 volts; otherwise remain unchanged. Volts; Dynamic voltage regulation coefficient Dimensionless, used for subsequent closed-loop control.

[0120] The output voltage of each electrode node in the deep electrode array is adjusted according to the dynamic voltage regulation coefficients until the deep electric field change parameters enter the preset stable range.

[0121] Next, the system can As a scaling factor, the deviation is expressed by the following formula. Execution ratio adjustment:

[0122] If 5 consecutive samples are taken after adjustment | If the value is ≤20 millivolts, the node is marked as "converged". When all nodes converge, the system determines that the deep electric field has converged, stops adjustment, and locks the output voltage.

[0123] The system collects a first time interval from the start of the gesture to the output of the gesture feature, and a second time interval from the start of the gesture to the output of the asynchronous accelerated sampling data.

[0124] The start time can be the sampling time when the impedance change rate of any node first exceeds 1000 ohms per millisecond. First time interval ,in, To correct the hand gesture characteristics Output time; second time interval ,in, To accelerate asynchronous sampling of data frames Full output time.

[0125] The ratio of the time difference between the first and second time intervals to the first time interval is determined as the identification delay compression value.

[0126] The system can be calculated using formulas. (Dimensionless) calculation yields the recognition delay compression value. .

[0127] Assuming the system measures =80 milliseconds, =48 milliseconds, then it can be calculated using the above formula. =0.4 indicates that the latency is compressed by 40%. It can be used to generate subsequent gesture recognition optimization parameters.

[0128] Therefore, according to the above implementation method, the system can dynamically adjust the voltage step size based on the compensation response time of each node during the finger approach stage, realize the rapid convergence of the deep electric field, and accurately quantify and identify the degree of delay compression by calculating the ratio of the first time interval to the second time interval, thus providing low-latency and highly stable performance indicators for touch interaction.

[0129] In some embodiments, clock alignment and voltage write-back operations are performed on multiple target processing units of the touchscreen according to gesture recognition optimization parameters to output gesture recognition results, including: The clock skew and voltage write-back values ​​are obtained by parsing the gesture recognition optimization parameters.

[0130] Among them, clock deviation Defined as: the difference between the measured delay from data input to output of each target processing unit and the unified reference delay, using a 2000 Hz master clock as a reference, expressed in milliseconds. The voltage write-back amount can be denoted as... This could refer to: parameters optimized by gesture recognition. middle The initial write-back voltage offset, obtained after normalization, in millivolts, is used to adjust the starting point for subsequent iterations.

[0131] Adjust the input data buffer capacity of each target processing unit according to the clock deviation to align the output times of each target processing unit.

[0132] Specifically, the system can increase or decrease the FIFO depth in the positive and negative directions of ΔCLK: if If so, increase the input buffer capacity of this unit to 256 sampling points and extend the dwell time; This reduces the number of sampling points to 128, shortening the dwell time; ultimately, the residual error of the output time of all units relative to the reference delay is ≤1 millisecond, completing the output time alignment (i.e., clock alignment).

[0133] Monitor the electric field fluctuation values ​​of the deep electric field change parameters of each target processing unit during operation.

[0134] Among them, electric field fluctuation value The definition can be: in 5 consecutive samples The difference between the maximum and minimum values, expressed in millivolts. For example, the system can read the values ​​of each node in real time at a frequency of 1000 Hz. It is used for stability determination.

[0135] If the electric field fluctuation value of any target processing unit exceeds the preset stability threshold, the compensation voltage value and the measured electric field deviation of that target processing unit are recorded.

[0136] Generally, the stability threshold can be taken as 20 millivolts; when When the voltage is >20 millivolts, the system can record the current node compensation voltage. Deviation from actual measurement , forming a record pair For iterative adjustment.

[0137] Based on the voltage write-back amount, the recorded compensation voltage values, and the measured deviations of each electric field, the output voltage of the corresponding electrode nodes in the deep electrode array is iteratively adjusted until the fluctuation values ​​of each electric field are lower than the preset stability threshold.

[0138] Specifically, the system can iteratively adjust the output voltage of the corresponding electrode nodes in the deep electrode array according to the following iterative formula:

[0139] The step size is adjustable in 5 millivolts; after each adjustment, the system waits 10 milliseconds before measuring Δ. If the voltage is ≤20 mV for 3 consecutive iterations, then the iteration is terminated and the circuit is locked. This achieves electric field convergence.

[0140] Gesture action features are extracted from asynchronous accelerated sampling data, and gesture recognition results are generated based on the comparison results of gesture action features with preset gesture action amplitude and preset gesture action frequency.

[0141] Specifically, the extraction method can follow the one described above. , The f-process yields the measured feature vector. .

[0142] Preset gesture range =100 millivolts, preset gesture frequency =8 Hz; the system can calculate the Euclidean distance using the following formula:

[0143] If D ≤ 25 millivolts, it is determined to be a "valid swipe gesture", and the recognition result is output as: {"Gesture Type": "Swipe", "Confidence Level": 1} When the confidence level is ≥0.75, the data is reported to the application layer.

[0144] Therefore, according to the above implementation method, the system can eliminate electric field fluctuations through voltage write-back iteration based on clock alignment, and quickly generate high-confidence gesture recognition results by means of preset feature comparison, so as to achieve low-latency and high-stability touch interaction.

[0145] Figure 2 This is a structural block diagram of a gesture recognition system for a touchscreen according to an embodiment of the present invention.

[0146] like Figure 2 As shown, the gesture recognition system of this touchscreen includes: The electric field distribution acquisition module 210 is used to acquire electric field distribution data on the surface of the touch screen when the user's finger is within a preset detection distance range of the touch screen. The electric field distribution data includes the deep electric field change parameters of the deep electrode array of the touch screen.

[0147] The gesture feature extraction module 220 is used to extract gesture action features from electric field distribution data and perform amplitude and frequency joint correction on the gesture action features according to a preset electric field strength benchmark.

[0148] The asynchronous sampling acceleration module 230 is used to perform asynchronous acceleration processing on the sampling frequencies of the shallow electrode array and the deep electrode array of the touch screen according to the compressed delay time, and to sample and obtain asynchronous accelerated sampling data; the compressed delay time is obtained by performing amplitude and frequency joint correction through gesture action features.

[0149] The delay compression metering module 240 is used to adjust the output voltage of the deep electrode array until the deep electric field change parameters converge according to the deep electric field change parameters in the asynchronous accelerated sampling data, and to determine the recognition delay compression value of the gesture action from the current touch screen according to the electric field strength reference.

[0150] The gesture optimization measurement module 250 is used to calculate gesture recognition optimization parameters based on the recognition delay compression value and the current asynchronous accelerated sampling data; the current asynchronous accelerated sampling data includes the converged deep electric field change parameters.

[0151] The recognition result generation module 260 is used to perform clock alignment and voltage write-back operations on multiple target processing units of the touch screen according to the gesture recognition optimization parameters, so as to output the gesture recognition result.

[0152] The specific functions and examples of each module and submodule of the device in this embodiment of the invention can be found in the relevant descriptions of the corresponding steps in the above method embodiments, and will not be repeated here.

[0153] According to embodiments of the present invention, the above-described method of the present invention can be applied to an electronic device and a readable storage medium.

[0154] Figure 3 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0155] like Figure 3 As shown, the electronic device 600 includes a computing unit 601, which can perform various appropriate actions and processes based on a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. The RAM 603 may also store various programs and data required for the operation of the electronic device 600. The computing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0156] Multiple components in electronic device 600 are connected to I / O interface 605, including: input unit 606, such as keyboard, mouse, etc.; output unit 607, such as various types of displays, speakers, etc.; storage unit 608, such as disk, optical disk, etc.; and communication unit 609, such as network card, modem, wireless transceiver, etc. Communication unit 609 allows electronic device 600 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0157] The computing unit 601 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as a gesture recognition method for a power distribution automation terminal touch screen. For example, in some embodiments, a gesture recognition method for a power distribution automation terminal touch screen can be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 600 via ROM 602 and / or communication unit 609. When the computer program is loaded into RAM 603 and executed by the computing unit 601, one or more steps of the gesture recognition method for a power distribution automation terminal touch screen described above can be performed. Alternatively, in other embodiments, the computing unit 601 may be configured by any other suitable means (e.g., by means of firmware) to perform a gesture recognition method for a power distribution automation terminal touchscreen.

[0158] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0159] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0160] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0161] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0162] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0163] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0164] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0165] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the principles of this invention should be included within the scope of protection of this invention.

Claims

1. A gesture recognition method for a power distribution automation terminal touch screen, characterized in that, include: When the user's finger is within a preset detection distance range of the touch screen, the electric field distribution data on the surface of the touch screen is collected. The electric field distribution data includes the deep electric field change parameters of the deep electrode array of the touch screen. Gesture action features are extracted from the electric field distribution data, and amplitude-frequency joint correction is performed on the gesture action features according to a preset electric field strength benchmark. Asynchronous acceleration processing is performed on the sampling frequencies of the shallow electrode array and the deep electrode array of the touch screen based on the compressed delay time, and asynchronous acceleration sampling data is obtained. The compressed delay time is obtained by performing amplitude and frequency joint correction on the gesture action features. The output voltage of the deep electrode array is adjusted according to the deep electric field change parameters in the asynchronous accelerated sampling data until the deep electric field change parameters converge, and the recognition delay compression value of the gesture action is determined from the current touch screen according to the electric field strength reference. The gesture recognition optimization parameters are calculated based on the recognition delay compression value and the current asynchronous accelerated sampling data, wherein the current asynchronous accelerated sampling data includes the converged deep electric field change parameters. Based on the gesture recognition optimization parameters, clock alignment and voltage write-back operations are performed on multiple target processing units of the touch screen to output gesture recognition results.

2. The method according to claim 1, characterized in that, The acquisition of electric field distribution data on the surface of the touch screen includes: Multiple first electric field intensity values ​​are obtained from each electrode node of the shallow electrode array, and a shallow reference electric field is constructed based on each of the first electric field intensity values. In response to the user's finger being within a preset detection distance range of the touchscreen, the impedance change rate of the shallow electrode array is obtained; If the impedance change rate exceeds the preset trigger condition value, then multiple second electric field strength values ​​are obtained from each electrode node of the shallow electrode array, and multiple third electric field strength values ​​are simultaneously obtained from each electrode node of the deep electrode array. Based on the shallow reference electric field, differential processing is performed on each of the third electric field intensity values ​​to generate the deep electric field variation parameters; The electric field distribution data is obtained by performing time alignment between each of the second electric field intensity values ​​and each of the third electric field intensity values ​​after differential processing.

3. The method according to claim 2, characterized in that, The extraction of gesture features from the electric field distribution data includes: A frequency domain transformation is performed on the electric field distribution data to obtain the signal-to-noise ratio values ​​of each electrode node; Electrode nodes with a signal-to-noise ratio higher than a preset signal-to-noise threshold are used as enhanced electric field signal channels. Filtering is performed on the signal components of each enhanced electric field signal channel, and an enhanced electric field signal set is constructed based on the filtered enhanced electric field signal channels. Weighting coefficients are calculated based on the ratio of each signal-to-noise ratio value to a preset reference signal-to-noise ratio value. Then, the corresponding enhanced electric field signal channels in the enhanced electric field signal set are weighted according to each weighting coefficient to obtain multiple weighted electric field signals. Normalization is performed on each of the weighted electric field signals to generate a set of electric field signal components; Envelope features, phase features, and periodic features are extracted from the set of electric field signal components. The maximum amplitude of the envelope feature is used as the gesture amplitude, the time offset of the phase feature is used as the gesture phase, and the reciprocal of the periodic feature is used as the gesture frequency to obtain the gesture feature.

4. The method according to claim 3, characterized in that, The step of performing amplitude-frequency joint correction on the gesture characteristics based on a preset electric field strength reference includes: Obtain a preset electric field strength reference, which includes the amplitude and frequency of a standard hand gesture. The first ratio of the gesture amplitude to the standard gesture amplitude is calculated, and the gesture amplitude is multiplied by the first ratio to obtain the corrected gesture amplitude. The second ratio of the gesture frequency to the standard gesture frequency is calculated, and the gesture frequency is multiplied by the second ratio to obtain the corrected gesture frequency. The corrected amplitude and the corrected frequency of the action are used as the corrected gesture action features to complete the joint correction of amplitude and frequency.

5. The method according to claim 1, characterized in that, The asynchronous acceleration processing of the sampling frequencies of the shallow electrode array and the deep electrode array of the touch screen based on the compressed delay time, and the sampling of asynchronous acceleration sampling data, includes: The delay level of each electrode node in the shallow electrode array and the deep electrode array is determined based on the compressed delay time. Electrode nodes in the shallow electrode array with a delay level lower than a preset threshold are identified as high-speed sampling channels, and electrode nodes in the deep electrode array with a delay level not lower than a preset delay threshold are identified as constant-speed sampling channels. Increase the sampling frequency of the high-speed sampling channel, maintain the sampling frequency of the constant-speed sampling channel, and reduce the sampling frequency of the remaining electrode nodes; Restart the shallow electrode array and the deep electrode array, and acquire asynchronous accelerated sampling data that is output in parallel within the same clock domain by the high-speed sampling channel and the constant-speed sampling channel.

6. The method according to claim 5, characterized in that, The step of adjusting the output voltage of the deep electrode array according to the deep electric field change parameters in the asynchronous accelerated sampling data until the deep electric field change parameters converge, and determining the recognition delay compression value of the gesture action from the current touch screen according to the electric field strength reference, includes: The compensation response time corresponding to the deep electric field change parameters of each electrode node in the deep electrode array is calculated based on the asynchronous accelerated sampling data. The voltage adjustment step size of each electrode node in the deep electrode array is adjusted according to the comparison results between each compensation response time and the preset response time interval, thereby generating a dynamic voltage adjustment coefficient for multiple electrode nodes. The output voltage of each electrode node in the deep electrode array is adjusted according to each of the dynamic voltage adjustment coefficients until the deep electric field change parameters enter the preset stable range. A first time interval is collected from the start time of the gesture to the output time of the gesture feature, and a second time interval is collected from the start time of the gesture to the output time of the asynchronous accelerated sampling data; The ratio of the time difference between the first time interval and the second time interval to the first time interval is determined as the recognition delay compression value.

7. The method according to claim 6, characterized in that, The step of performing clock alignment and voltage write-back operations on multiple target processing units of the touchscreen according to the gesture recognition optimization parameters to output gesture recognition results includes: The clock skew and voltage write-back values ​​are obtained by parsing the gesture recognition optimization parameters. Adjust the input data buffer capacity of each target processing unit according to the clock deviation to align the output times of each target processing unit. Monitor the electric field fluctuation values ​​of the deep electric field change parameters of each of the target processing units during operation; If the electric field fluctuation value of any of the target processing units exceeds the preset stability threshold, the compensation voltage value and the measured electric field deviation of the target processing unit are recorded. Based on the voltage write-back amount, the recorded compensation voltage values, and the measured electric field deviations, the output voltage of the corresponding electrode nodes in the deep electrode array is iteratively adjusted until each electric field fluctuation value is lower than a preset stability threshold. Gesture action features are extracted from the asynchronous accelerated sampling data, and the gesture recognition result is generated based on the comparison results of the gesture action features with preset gesture action amplitude and preset gesture action frequency.

8. A gesture recognition system for a touchscreen, characterized in that, include: An electric field distribution acquisition module is used to acquire electric field distribution data on the surface of the touch screen when the user's finger is within a preset detection distance range of the touch screen. The electric field distribution data includes the deep electric field change parameters of the deep electrode array of the touch screen. The gesture feature extraction module is used to extract gesture action features from the electric field distribution data, and to perform amplitude-frequency joint correction on the gesture action features according to a preset electric field strength benchmark. An asynchronous sampling acceleration module is used to perform asynchronous acceleration processing on the sampling frequencies of the shallow electrode array and the deep electrode array of the touch screen according to the compressed delay time, and to sample asynchronous accelerated sampling data. The compressed delay time is obtained by performing amplitude and frequency joint correction on the gesture action features. The delay compression metering module is used to adjust the output voltage of the deep electrode array until the deep electric field change parameters converge according to the deep electric field change parameters in the asynchronous accelerated sampling data, and to determine the recognition delay compression value of the gesture action from the current touch screen according to the electric field strength reference. The gesture optimization measurement module is used to calculate gesture recognition optimization parameters based on the recognition delay compression value and the current asynchronous accelerated sampling data, wherein the current asynchronous accelerated sampling data includes converged deep electric field change parameters. The recognition result generation module is used to perform clock alignment and voltage write-back operations on multiple target processing units of the touch screen according to the gesture recognition optimization parameters, so as to output the gesture recognition result.

9. An electronic device, characterized in that, include: At least one processor; as well as The memory that is communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that, Computer instructions are used to cause a computer to perform the method according to any one of claims 1-7.

Images