Capacitive touch screen input event classification and processing system

Abstract

The invention relates to the technical field of sensing, and discloses a capacitive touch screen input event classification and processing system which comprises an input capture module, an event classification module, an intention prediction module, a dynamic response module, an energy efficiency optimization module and a multi-channel cooperation module. When an input event of the capacitive touch screen is identified, a dynamic noise suppression model is established, and adaptive filtering strategies are configured for different use scenes, so that the reliability of signal analysis in various electromagnetic interference environments is ensured, and meanwhile, the spatial and temporal distribution characteristics of capacitance variation are analyzed in real time; real touch operation and environmental electromagnetic noise can be effectively distinguished, the identification fidelity of basic operations such as clicking and sliding is ensured, the false triggering rate is reduced, the operation accuracy in a weak signal scene is improved, and when multi-finger operation is detected, the consistency of contact coordinates in a curved surface state and a plane state is maintained, so that the accuracy of the touch operation is improved. And the track continuity of writing and drawing type fine operation is ensured.

Description

Technical Field

[0001] The present invention relates to the field of sensing technology, and in particular to a capacitive touch screen input event classification and processing system. Background Art

[0002] Sensing technology is a multidisciplinary modern science and engineering technology that is about acquiring information from natural sources and processing and identifying it. It involves the planning, design, development, manufacturing, testing, application, and evaluation and improvement activities of sensors, information processing and identification.

[0003] The capacitive touch screen input event classification and processing system is a core human-computer interaction platform that integrates artificial intelligence and new sensing technologies. The system achieves high-precision touch response by real-time analysis of capacitance change patterns. It is a key interactive component of modern smart terminals, responsible for converting user touch operations into machine instructions. The multi-channel capacitive sensing matrix can capture contact changes to ensure interaction accuracy in complex environments.

[0004] At present, due to the strong electromagnetic interference in the use scenarios of mobile devices, the built-in filtering algorithm of the device has limited ability to suppress sudden noise when performing contact signal recognition, and cannot distinguish between effective touch and environmental electromagnetic noise in real time. When there is charger coupling interference and environmental RF noise, the false trigger rate of the touch will increase, reducing the reliability of the touch. At the same time, on curved screen and folding screen devices, the edge contact coordinate offset caused by the screen curvature cannot be dynamically corrected, which will cause the edge sliding trajectory to be broken and multi-touch association errors, and there is a lack of real-time geometric calibration mechanism when contact drift occurs. When identifying three-dimensional touch and pressure operations, since the traditional system uses a fixed pressure threshold segmentation, it cannot adaptively distinguish the operation force characteristics of different users, resulting in an increase in the pressure grading misjudgment rate, further affecting the interaction accuracy of scenarios such as handwriting input and game control.

[0005] Therefore, a capacitive touch screen input event classification and processing system is proposed to solve the above problems. Summary of the Invention

[0006] (1) Technical problems solved In view of the shortcomings of the existing technology, the present invention provides a capacitive touch screen input event classification and processing system, which solves the problems raised in the above background technology.

[0007] (2) Technical solution To achieve the above objectives, the present invention provides the following technical solutions: a capacitive touch screen input event classification and processing system, comprising: Input capture module: collects raw capacitance change matrix data of the capacitive touch screen in real time and generates an input trigger signal when it detects that the capacitance change exceeds the noise threshold; Event classification module: This module extracts features from the capacitance change matrix using a deep convolutional neural network and classifies input events into at least one of the following types: click, slide, long press, multi-finger zoom, and hover operation through spatiotemporal feature fusion analysis. Intent Prediction Module: Builds a user behavior profile based on historical interaction behavior data. Combined with the type, trajectory, and pressure distribution of current input events, it predicts the user's operation intention and generates an intent confidence score. Dynamic response module: Adaptively adjusts touch screen sampling frequency, touch point recognition accuracy, and interface response strategy based on intent confidence score and current system operation status; Energy efficiency optimization module: By monitoring the temporal distribution of input events and interface interaction hotspots, it dynamically shuts down sensor circuits in inactive areas to reduce system power consumption; Multi-channel collaboration module: When multi-finger operation is recognized, a cross-channel touch point association model is established to eliminate the touch point coordinate drift error caused by screen edge effects.

[0008] Preferably, the noise threshold is set as follows: Continuously collect background capacitance value matrix in no-touch state ; Calculate the current capacitance matrix and The Euclidean distance of: in, is the background capacitance reference matrix in the no-touch state, is the real-time capacitance value of the i-th sensing unit; when Preset noise margin , it is determined to be a valid input trigger.

[0009] Preferably, the spatiotemporal feature fusion analysis includes: Divide the capacitance variation matrix into subframe blocks according to the time series; Extract spatial gradient features for each subframe block and time rate of change characteristics ; The event type probability is calculated using the feature weighted fusion formula: in, is the spatial gradient feature, is the time rate of change, , is the dynamic weighting coefficient.

[0010] Preferably, the method for eliminating the contact coordinate drift error is: Detect the touch point coordinates set in the edge area of ​​the screen ; Constructing a central area reference point set Calculate the coordinate correction matrix through the affine transformation model: The parameters By minimizing Solve, for the affine transformation matrix parameters, , is the translation vector.

[0011] Preferably, the intention prediction module includes: Establish a user behavior prediction model based on LSTM network. The input features include: Similarity between the operation time and the same period in history ; Contact pressure distribution entropy ; Track curvature change ; The output dimensions include the probability distribution of three core intents: application launch, page jump, and content zoom.

[0012] Preferably, the strategy generation logic of the dynamic response module is: When the intent confidence score is greater than 0.9, the preloading mechanism is enabled to render the target interface in advance; When the system battery level is less than 20%, the touch point recognition accuracy is reduced from sub-pixel level to pixel level; Force the 120Hz full-area sampling mode to be enabled in gaming scenarios.

[0013] Preferably, the implementation of the energy efficiency optimization module includes: Divide the screen into A dynamic perception block; Statistics of each block in the time window Trigger frequency within ; when When , the sensing circuit of the block is turned off; When a cross-block sliding operation is detected, the adjacent blocks are woken up in advance according to the predicted path of the sliding direction.

[0014] Preferably, it also includes: Pressure state recognition submodule: It identifies three states: light pressure, heavy pressure, and suspension through the gradient distribution pattern of capacitance change. The suspension state determination conditions are: in, is the second-order partial derivative of capacitance change in the X direction, is the suspension determination threshold.

[0015] Preferably, the identification process of light pressure and heavy pressure is: Calculate the average capacitance change in the core area of ​​the contact ; Calculate the capacitance variation variance in the peripheral transition region ; when If , it is judged as heavy pressure, otherwise it is light pressure.

[0016] Preferably, the system integrates a quantization processing unit, Encoding the capacitance variation matrix as a quantum state ; Extract features through quantum convolution operation: The Hamiltonian Built from trainable parameterized quantum circuits, is the unitary evolution time.

[0017] (3) Beneficial effects Compared with the prior art, the present invention provides a capacitive touch screen input event classification and processing system, which has the following beneficial effects: 1. In the present invention, by providing an intelligent noise reduction unit, a dynamic noise suppression model is established when recognizing capacitive touch screen input events, and an adaptive filtering strategy is configured for different usage scenarios to ensure the reliability of signal analysis in various electromagnetic interference environments. At the same time, the spatiotemporal distribution characteristics of the capacitance change are analyzed in real time, which can effectively distinguish between real touch operations and environmental electromagnetic noise, ensure the recognition fidelity of basic operations such as clicking and sliding, reduce the false trigger rate, and improve the operation accuracy in weak signal scenarios.

[0018] 2. In the present invention, by setting a deformation compensation core, when positioning the touch points of curved screen and folding screen devices, the coordinate distortion parameters caused by the screen curvature are dynamically calculated, and the geometric mapping relationship between the edge touch points and the center area is corrected in real time, so that the system automatically eliminates the touch point drift caused by screen deformation. When multi-finger operation is detected, the spatial position of the touch points is instantly corrected by establishing a cross-region touch point topology association model, maintaining the consistency of the touch point coordinates in the curved and flat states, and ensuring the trajectory continuity of fine operations such as writing and painting.

[0019] 3. In the present invention, by setting up a behavior analysis engine, when performing three-dimensional touch and pressure operation recognition, the capacitance gradient distribution and pressure conduction pattern of the contact area are simultaneously deconstructed, and the multi-dimensional states of light pressure, heavy pressure and floating operations are judged in real time. The operation force grading standards are dynamically matched according to different application scenarios, so that the system can accurately distinguish differentiated interactive intentions such as game control and text input, reduce the pressure grading misjudgment caused by traditional fixed thresholds, and enhance the accuracy of operation semantic understanding in complex human-computer interaction scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the overall system architecture of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0022] See also Figure 1 , a capacitive touch screen input event classification and processing system, comprising: Input capture module: collects raw capacitance change matrix data of the capacitive touch screen in real time and generates an input trigger signal when it detects that the capacitance change exceeds the noise threshold; Event classification module: This module extracts features from the capacitance change matrix using a deep convolutional neural network and classifies input events into at least one of the following types: click, slide, long press, multi-finger zoom, and hover operation through spatiotemporal feature fusion analysis. Intent Prediction Module: Builds a user behavior profile based on historical interaction behavior data. Combined with the type, trajectory, and pressure distribution of current input events, it predicts the user's operation intention and generates an intent confidence score. Dynamic response module: Adaptively adjusts touch screen sampling frequency, touch point recognition accuracy, and interface response strategy based on intent confidence score and current system operation status; Energy efficiency optimization module: By monitoring the temporal distribution of input events and interface interaction hotspots, it dynamically shuts down sensor circuits in inactive areas to reduce system power consumption; Multi-channel collaboration module: When multi-finger operation is recognized, a cross-channel touch point association model is established to eliminate touch point coordinate drift errors caused by screen edge effects; The process of setting the noise threshold is: Continuously collect background capacitance value matrix in no-touch state ; Calculate the current capacitance matrix and The Euclidean distance of: in, is the background capacitance reference matrix in the no-touch state, is the real-time capacitance value of the i-th sensing unit; when Preset noise margin When , it is determined to be a valid input trigger; Spatiotemporal feature fusion analysis includes: Divide the capacitance variation matrix into subframe blocks according to the time series; Extract spatial gradient features for each subframe block and time rate of change characteristics ; The event type probability is calculated using the feature weighted fusion formula: in, is the spatial gradient feature, is the time rate of change, , is the dynamic weighting coefficient; The method to eliminate the contact coordinate drift error is: Detect the touch point coordinates set in the edge area of ​​the screen ; Constructing a central area reference point set Calculate the coordinate correction matrix through the affine transformation model: The parameters By minimizing Solve, for the affine transformation matrix parameters, , is the translation vector; The intent prediction module includes: Establish a user behavior prediction model based on LSTM network. The input features include: Similarity between the operation time and the same period in history ; Contact pressure distribution entropy ; Track curvature change ; The output dimensions include the probability distribution of three core intents: application launch, page jump, and content zoom; The strategy generation logic of the dynamic response module is: When the intent confidence score is greater than 0.9, the preloading mechanism is enabled to render the target interface in advance; When the system battery level is less than 20%, the touch point recognition accuracy is reduced from sub-pixel level to pixel level; Force the 120Hz full-area sampling mode to be enabled in gaming scenarios; The implementation methods of the energy efficiency optimization module include: Divide the screen into A dynamic perception block; Statistics of each block in the time window Trigger frequency within ; when When , the sensing circuit of the block is turned off; When a cross-block sliding operation is detected, the adjacent blocks are woken up in advance according to the predicted path of the sliding direction; Also includes: Pressure state recognition submodule: It identifies three states: light pressure, heavy pressure, and suspension through the gradient distribution pattern of capacitance change. The suspension state determination conditions are: in, is the second-order partial derivative of capacitance change in the X direction, is the suspension determination threshold; The identification process of light pressure and heavy pressure is as follows: Calculate the average capacitance change in the core area of ​​the contact ; Calculate the capacitance variation variance in the peripheral transition region ; when When , it is judged as heavy pressure, otherwise it is light pressure; The system integrates a quantized processing unit. Encoding the capacitance variation matrix as a quantum state ; Extract features through quantum convolution operation: The Hamiltonian Built from trainable parameterized quantum circuits, is the unitary evolution time.

[0023] Example 1: Implementing full-process processing from raw signal acquisition to intelligent response through multi-module collaboration First, the hardware environment is deployed: a high-precision capacitive sensor chip is integrated into the driver layer of the capacitive touch screen. Its sampling frequency is configured to 240Hz to capture signal changes. At the same time, it is equipped with a dedicated AI coprocessor to run deep learning models in real time. After the system starts, initialization calibration is performed: the environmental background noise is continuously collected in the non-touch state, and the noise basis matrix is ​​dynamically generated and stored in the Flash memory. This matrix will serve as a reference for subsequent signal processing.

[0024] When a user's finger touches the screen, the sensor chip generates a real-time distribution map of the capacitance change. The signal preprocessing unit first calculates the Euclidean distance between the current frame and the noise floor. If a mutation value in a specific region exceeds a preset tolerance, the event classification process is immediately triggered, and the spatiotemporal feature extraction engine begins its work. In the spatial dimension, the Sobel operator is used to calculate the gradient distribution pattern of the contact area and identify the contact's morphological contour. In the temporal dimension, a sliding window is used to compare the rate of change of five consecutive frames of data to construct a trajectory dynamic vector. These two types of features are weighted and fused and then fed into a pre-trained MobileNetV3 classification model, which has been trained on millions of samples and can output recognition results for five types of operations: click, slide, long press, multi-finger zoom, and hover.

[0025] For operations recognized as sliding, the deformation compensation module is immediately activated. On curved screen devices, the system automatically loads the screen curvature parameters and reconstructs the touch coordinate mapping relationship through the affine transformation matrix: the physical touch coordinates in the edge area are mapped to the virtual plane coordinate system, reducing the pixel positioning deviation caused by refraction of the glass cover. When multi-finger operation is detected, the topological association algorithm will establish a set of differential equations for the touch motion trajectory, predict the touch motion trend, and wake up the sensor circuit in the target area in advance to reduce response delay.

[0026] At the intent prediction level, the user behavior analysis engine continuously records operating habits. When it detects that the standard deviation of the handwriting pressure value in the handwriting input scenario is lower than 0.8 for five consecutive times, it determines that the user is in fine operation mode and automatically lowers the pressure classification threshold from the default 1.2N to 0.6N. When three consecutive 120° sharp turns are detected in the game scene, the preloading mechanism is activated to render the map resources in advance. All operation data is encrypted with AES-256 and uploaded to the edge computing node, realizing online incremental learning of model parameters while ensuring privacy and security.

[0027] The energy efficiency optimization subsystem reduces power consumption through dynamic block management, dividing the screen into 12×24 independently controllable power units. When a specific block has no effective touch for 30 consecutive seconds, the sensor circuit automatically switches to microampere-level sleep mode. When a horizontal sliding operation is detected, the power management unit wakes up the sensor units of the first six blocks according to the sliding direction gradient, reducing the system standby power consumption to 3.2mW.

[0028] For the deployment of quantum computing units, on flagship devices equipped with quantum processors, the system encodes the capacitance matrix into a 16-qubit state, performs convolution operations through parameterized quantum circuits, completes feature extraction within 3μs, and the output results are input into the classical AI model after quantum measurement to complete intent classification, forming a hybrid computing paradigm.

[0029] A fault recovery mechanism ensures system robustness: when a persistent abnormality in single-point capacitance is detected, the system automatically switches to the backup sensing channel and sends sensor calibration parameters through the cloud-based diagnostic system. All core modules adopt a dual-DSP redundant architecture to achieve industrial-grade operational reliability.

[0030] Example 2: Vehicle-mounted intelligent cockpit application example This embodiment uses the smart cockpit of a new energy vehicle as an application scenario to describe in detail the implementation process of the capacitive touch screen input event classification and processing system.

[0031] The system is deployed on the large central control screen in the vehicle, focusing on solving safe interaction problems in driving scenarios. The hardware configuration adopts an automotive-grade triple-redundant architecture: the main control unit is equipped with a multi-core processor to realize parallel signal processing, the sensing layer adopts a double-layer mutual capacitance matrix, and integrates a radar-assisted positioning module. When the system starts, it performs vehicle environment self-calibration: the vehicle driving status data is collected through the vibration sensor, and the vibration noise compensation curve is dynamically generated and stored in the EEPROM.

[0032] When the driver's finger touches the central control screen, the system prioritizes activating the safe interaction mode. The signal acquisition unit synchronously processes the capacitance change and positioning data, and reduces the contact drift caused by vehicle bumps through the spatiotemporal alignment algorithm. The event classification module loads the vehicle-specific model and adds a 200ms delay confirmation mechanism when identifying click operations to avoid false triggering caused by driving bumps. For sliding operations, the steering wheel angle sensor data is combined to predict the operation intention. When sliding the map to the left while turning right is detected, it is automatically determined as a map browsing need rather than an operation instruction.

[0033] The curved screen correction module is specially optimized for the in-vehicle environment: in direct sunlight scenarios, it dynamically improves touch sensitivity based on light sensor data to compensate for capacitive signal attenuation caused by strong light. In low-temperature environments, the deformation compensation algorithm is automatically enabled, and the coordinate mapping model is reconstructed by loading the screen's thermal expansion coefficient parameters. When processing multi-finger operations, the system prioritizes locking the touch permissions of the co-pilot area. When the vehicle speed is greater than 60km / h, complex gesture operations of the entertainment system are automatically blocked.

[0034] The intention prediction engine integrates driving behavior analysis: when it detects three consecutive map zoom operations accompanied by brake pedal signals, it preloads the 3D panoramic model of the navigation route. When the air conditioning temperature adjustment operation is recognized, it combines the cabin temperature and humidity sensors and passenger location data to automatically generate a zone temperature control recommendation plan. All sensitive operations must be biometrically reviewed by the DMS driver monitoring system to ensure driving safety.

[0035] The energy efficiency management system uses dynamic voltage regulation technology: it enables high-performance mode when the vehicle is charging, switches to energy-saving mode when powered by the battery, and reduces the voltage of the sensor circuit in non-driving areas to 1.8V. The system monitors the distribution of touch hotspots in real time. When the vehicle enters highway cruising mode, it automatically shuts down the sensor units outside the air-conditioning control area, reducing system power consumption to 15mW.

[0036] The fault-tolerant mechanism achieves ASIL-D functional safety: when the capacitive sensing unit detects a persistent abnormal signal, it immediately switches to millimeter-wave radar single-mode positioning mode and displays a warning message through the HUD. The system performs a self-check every 500ms and stores triple backup of key data to ensure operational reliability in extreme vibration environments.

[0037] Innovations in technical elements: Enhanced driving safety: Operation delay confirmation mechanism avoids accidental touches due to bumps, DMS biometric verification of sensitive commands, vehicle speed-linked touch permission classification, environmental adaptation: sensitivity compensation for direct sunlight scenes, real-time correction of low-temperature deformation, dual-mode power consumption management in charge and discharge states, cross-system collaboration: steering wheel angle prediction of operating intentions, brake signal triggers navigation pre-loading, temperature and humidity data optimizes air conditioning control, safety redundant architecture: millimeter-wave radar backup positioning channel, triple data storage backup, ASIL-D level self-test mechanism.

[0038] Example 3: Application of smart home central control panel This embodiment focuses on smart home scenarios and is deployed on wall-embedded capacitive touch panels, focusing on solving the security control and energy-saving interaction problems in a multi-user environment.

[0039] The hardware adopts an ultra-thin flexible screen design and integrates a distributed pressure sensor array. The main control unit is equipped with a Renesas RA8M1 MCU and a built-in privacy computing module. Multi-user registration is performed during system initialization: the user's fingertip capacitance feature map is collected through a special conductive mesh film, and a 512-dimensional biometric feature vector is generated and encrypted and stored in a secure chip to establish a personalized touch behavior model library.

[0040] When a user operates the control panel, the identity authentication module is activated first: by comparing the capacitance distribution pattern of the contact area with the pre-stored feature vector in real time, it automatically identifies the permissions of administrators and ordinary users. The event classification module loads a dedicated model for home scenarios and introduces a contact area verification mechanism when identifying click operations. When a child user attempts to operate the HVAC system, a contact area of ​​less than 40mm² will trigger an operation interception. For slow sliding of elderly users, the response window is automatically extended and visual feedback is enhanced.

[0041] In view of the physical characteristics of the deformable panel, the dynamic deformation compensation system works in real time: the flat touch mode is enabled when the panel is unfolded, and the surface algorithm is automatically switched when it is folded and stored. The coordinate mapping is reconstructed through the pre-loaded Bezier surface parameters to reduce the contact offset error in the folded state. When multiple users operate concurrently, the spatiotemporal conflict resolution algorithm automatically coordinates based on the operation priority: when the kitchen light adjustment and security deployment are triggered at the same time, the system responds to the security command first and delays the lighting operation for 5 seconds.

[0042] The energy-saving management system innovatively integrates environmental perception: it monitors the panel microenvironment through built-in temperature and humidity sensors. When the risk of condensation on the panel surface is detected, it automatically shuts down the sensing circuit in the bottom 20% area to prevent false triggering. In night mode, it enables low-light interaction, lowers the touch response threshold, and enables infrared assisted positioning. If the panel is idle for 15 minutes, the system enters deep sleep, leaving only the four corners active for wake-up detection.

[0043] The fault emergency mechanism includes triple protection: when water intrusion is detected, the hydrophobic coating electrode switching scheme is automatically activated; when encountering strong electromagnetic interference, it instantly switches to differential signal acquisition mode; in the event of physical damage, the self-repair circuit rebuilds the signal path through programmable fuses to maintain basic touch function.

[0044] Analysis of core technological innovations: Variable form adaptation: real-time mapping of Bezier surface parameters, seamless switching between folded and unfolded states, dynamic elimination of 1.8mm offset error, home scene safety protection: children's contact area interception mechanism, enhanced feedback for slow operations of the elderly, active defense against environmental risks, multi-user collaborative management: biocapacitive feature authentication, dynamic sorting of instruction priorities, local execution of privacy computing, ultra-low power architecture: low-light interaction mode, four-corner wake-up detection area, and selective shutdown of condensation areas.

[0045] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0046] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A capacitive touch screen input event classification and processing system, characterized by: include: Input capture module: collects raw capacitance change matrix data of the capacitive touch screen in real time and generates an input trigger signal when it detects that the capacitance change exceeds the noise threshold; Event classification module: This module extracts features from the capacitance change matrix using a deep convolutional neural network and classifies input events into at least one of the following types: click, slide, long press, multi-finger zoom, and hover operation through spatiotemporal feature fusion analysis. Intent Prediction Module: Builds a user behavior profile based on historical interaction behavior data. Combined with the type, trajectory, and pressure distribution of current input events, it predicts the user's operation intention and generates an intent confidence score. Dynamic response module: Adaptively adjusts touch screen sampling frequency, touch point recognition accuracy, and interface response strategy based on intent confidence score and current system operation status; Energy efficiency optimization module: By monitoring the temporal distribution of input events and interface interaction hotspots, it dynamically shuts down sensor circuits in inactive areas to reduce system power consumption; Multi-channel collaboration module: When multi-finger operation is recognized, a cross-channel touch point association model is established to eliminate the touch point coordinate drift error caused by screen edge effects.

2. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The noise threshold setting process is as follows: Continuously collect background capacitance value matrix in no-touch state ; Calculate the current capacitance matrix and The Euclidean distance of: in, is the background capacitance reference matrix in the no-touch state, is the real-time capacitance value of the i-th sensing unit; when Preset noise margin , it is determined to be a valid input trigger.

3. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The spatiotemporal feature fusion analysis includes: Divide the capacitance variation matrix into subframe blocks according to the time series; Extract spatial gradient features for each subframe block and time rate of change characteristics ; The event type probability is calculated using the feature weighted fusion formula: in, is the spatial gradient feature, is the time rate of change, , is the dynamic weighting coefficient.

4. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The method for eliminating the contact coordinate drift error is: Detect the touch point coordinates set in the edge area of ​​the screen ; Constructing a central area reference point set Calculate the coordinate correction matrix through the affine transformation model: The parameters By minimizing Solve, for the affine transformation matrix parameters, , is the translation vector.

5. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The intention prediction module includes: Establish a user behavior prediction model based on LSTM network. The input features include: Similarity between the operation time and the same period in history ; Contact pressure distribution entropy ; Track curvature change ; The output dimensions include the probability distribution of three core intents: application launch, page jump, and content zoom.

6. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The strategy generation logic of the dynamic response module is: When the intent confidence score is greater than 0.9, the preloading mechanism is enabled to render the target interface in advance; When the system battery level is less than 20%, the touch point recognition accuracy is reduced from sub-pixel level to pixel level; Force the 120Hz full-area sampling mode to be enabled in gaming scenarios.

7. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The implementation of the energy efficiency optimization module includes: Divide the screen into A dynamic perception block; Statistics of each block in the time window Trigger frequency within ; when When , the sensing circuit of the block is turned off; When a cross-block sliding operation is detected, the adjacent blocks are woken up in advance according to the predicted path of the sliding direction.

8. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: Also includes: Pressure state recognition submodule: It identifies three states: light pressure, heavy pressure, and suspension through the gradient distribution pattern of capacitance change. The suspension state determination conditions are: in, is the second-order partial derivative of capacitance change in the X direction, is the suspension determination threshold.

9. The capacitive touch screen input event classification and processing system according to claim 8, characterized in that: The identification process of light pressure and heavy pressure is as follows: Calculate the average capacitance change in the core area of ​​the contact ; Calculate the capacitance variation variance in the peripheral transition region ; when If , it is judged as heavy pressure, otherwise it is light pressure.

10. The capacitive touch screen input event classification and processing system according to claim 1, characterized in that: The system integrates a quantized processing unit. Encoding the capacitance variation matrix as a quantum state ; Extract features through quantum convolution operation: The Hamiltonian Built from trainable parameterized quantum circuits, is the unitary evolution time.

Images