Foreign matter detection and positioning system for touch screen based on capacitance matrix spectrum analysis

Abstract

The present application belongs to the technical field of capacitive sensing and signal processing, and relates to a foreign matter detection and positioning system for a touch screen based on capacitive matrix spectrum analysis. The system comprises: a sweep excitation signal generation module for generating a sweep excitation signal sequence and a reference clock reference; an electrode driving and network forming module for polling driving a transmitting electrode to establish an equivalent impedance network; a phase-sensitive demodulation module for extracting in-phase and quadrature component signals; a complex impedance matrix construction module for constructing a complex impedance characteristic matrix; a geometric feature extraction module for extracting local geometric mapping features; a substance classification module for determining the type of a real substance; a coordinate calculation module for measuring and calculating a two-dimensional space coordinate; and an abnormality report generation module for generating a structured abnormality report including a qualitative and positioning composite attribute. The present application solves the problem that, due to the lack of high-resolution wide-frequency-response data, the qualitative determination of the type of foreign matter on a touch screen and the spatial positioning accuracy need to be improved.

Description

Technical Field

[0001] This invention belongs to the technical field of capacitive sensing and signal processing, and relates to a foreign object detection and positioning system for touch screens based on capacitance matrix spectrum analysis. Background Technology

[0002] In capacitive touchscreen technology, accurately detecting and identifying foreign objects unexpectedly adhering to the screen surface remains a persistent technical challenge. These foreign objects, such as liquids, food residue, or metal particles, can alter the local capacitance distribution of the touchscreen's electrode network, potentially being misinterpreted by the system as valid touch input, leading to erroneous operations or causing touch response failure in localized areas. Therefore, developing detection technologies that can effectively distinguish between normal finger touches and foreign object interference directly contributes to improving the reliability of touchscreens in complex environments.

[0003] Currently, a common approach in the industry for handling such problems is based on capacitance threshold detection at a single operating frequency. This method monitors the capacitance change at each sensing node, triggering an anomaly flag when the change exceeds a preset threshold. Some improved solutions add simple analysis of the area or shape of the abnormal region to distinguish between small fingertip touches and large areas of hand or foreign object coverage. Other solutions employ switching measurements between a few discrete frequency points, attempting to use the capacitance differences that different substances may exhibit at different frequencies for preliminary classification. These methods can, to some extent, address specific, characteristic interferences.

[0004] However, the aforementioned traditional methods have limitations when dealing with diverse foreign object types and complex environmental changes. Relying solely on capacitance thresholds at a single frequency makes it difficult to distinguish substances with similar dielectric properties but completely different physical properties, such as water and certain types of gels. Furthermore, the detection threshold is susceptible to drift in reference capacitance caused by changes in ambient temperature and humidity, potentially leading to insufficient detection stability. Measuring using a few discrete frequencies results in limited resolution of the acquired spectral information. For many substances that exhibit characteristic differences only over a wide spectral range, this method provides insufficient data dimensionality to support accurate qualitative analysis. In these specific and complex scenarios, traditional solutions require improvement in high-resolution substance identification and precise shielding. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis.

[0006] A foreign object detection and localization system for touch screens based on capacitance matrix spectrum analysis includes: The frequency sweep excitation signal generation module generates a continuous frequency sweep excitation signal sequence covering a predetermined bandwidth according to the environmental monitoring instructions, and synchronously outputs the corresponding reference clock reference. The electrode driving and network forming module is used to perform time-sequential polling driving of the transmitting electrodes of the capacitive touch screen using a continuous frequency sweep excitation signal sequence, so as to establish a local equivalent impedance network that reflects the physical state of the surface of the capacitive touch screen. The phase-sensitive demodulation module is used to capture the analog signal modulated by the local equivalent impedance network and perform phase-sensitive demodulation on the analog signal based on the reference clock reference to extract the in-phase component signal and the quadrature component signal. The complex impedance matrix construction module calculates complex plane coordinate data based on in-phase and quadrature component signals, and combines the complex plane coordinate data to construct a complex impedance characteristic matrix. The geometric feature extraction module is used to process the complex impedance feature matrix to filter out global environmental changes and extract the local geometric mapping features of corresponding abrupt regions; The material classification module is used to compare local geometric mapping features with a pre-calibrated database of foreign object features to determine the type of substance in the mutation region. The coordinate calculation module is used to trace back the abnormal channel identifiers that contribute to the local geometric mapping features, and to use hardware array projection to calculate the two-dimensional spatial coordinates of the material type. The anomaly report generation module is used to stitch together the type of substance and two-dimensional spatial coordinates to generate a structured anomaly report that includes both qualitative and locational composite attributes.

[0007] A further aspect of the present invention includes a frequency sweep excitation signal generation module, which performs the following operations: Parse environmental monitoring commands to obtain preset scan parameter configuration files; Adjust the bias voltage of the voltage-controlled oscillator unit according to the preset scanning parameter configuration file to drive the voltage-controlled oscillator unit to output a continuous frequency sweep excitation signal sequence; The phase and frequency parameters of the continuous sweep excitation signal sequence are extracted and input into a high-precision phase-locked loop to generate a reference clock that is synchronized with the continuous sweep excitation signal sequence.

[0008] In a further embodiment of the present invention, an electrode driving and network forming module is used to perform the following steps: A scan timing generator is driven using a reference clock reference to generate array scan timing signals; The array scanning timing signal controls the analog multiplexed switch array to sequentially turn on the emitter electrodes of each row of the capacitive touch screen. A continuous frequency sweep excitation signal sequence is injected into the currently conducting transmitting electrode. The dielectric coupling effect of the screen surface attachments on the edge electric field is used to establish a local equivalent impedance network at the spatial intersection of the transmitting and receiving electrodes.

[0009] A further embodiment of the present invention includes a phase-sensitive demodulation module, which is used to perform the following operations: The analog signal and the reference clock reference are input to the analog multiplier for multiplication to generate a mixed signal; The mixed signal is input into a filter to remove high-frequency harmonic components and obtain low-frequency DC components; The low-frequency DC component is separated to extract the in-phase component signal representing the real part of the complex impedance and the quadrature component signal representing the imaginary part of the complex impedance.

[0010] A further aspect of the present invention includes a complex impedance matrix construction module, used to perform the following operations: Using the values ​​of the in-phase component signal as the real axis coordinates and the values ​​of the quadrature component signal as the imaginary axis coordinates, the complex impedance point of a single induction node at a specific frequency is calculated. Multiple complex impedance points measured at all scanning frequencies by the same sensing node are connected in ascending order of frequency to generate a continuous impedance trajectory array. Traverse all sensing nodes across the entire screen and combine all continuous impedance trajectory arrays to generate a complex impedance characteristic matrix.

[0011] A further aspect of the present invention includes a geometric feature extraction module, which performs the following operations: Calculate the squared Euclidean distance between the complex impedance characteristic matrix and the preset cleanroom reference matrix at each frequency point to form an Euclidean distance evolution sequence; Perform a first-order difference operation on the Euclidean distance evolution sequence to eliminate the overall linear translation of the trajectory caused by changes in environmental temperature and humidity; Based on the processed trajectory data, the resonant peak and valley points are located, and the radius of curvature and the tilt angles of the major and minor semi-axis at the resonant peak and valley points are calculated to form local geometric mapping features.

[0012] A further aspect of the present invention includes a material classification module, used to perform the following operations: The nearest neighbor pattern recognition algorithm is used to calculate the error distance between the local geometric mapping features and the standard feature vectors in the foreign object feature database; The minimum value in the error distance is compared with the preset error tolerance threshold. When the minimum value is less than the preset error tolerance threshold, the standard feature vector corresponding to the minimum value is determined as the matching result, and the circuit distribution attribute corresponding to the matching result is retrieved. The circuit distribution properties are analyzed to reconstruct the equivalent topological physical model, and the types of physical matter mapped by the equivalent topological physical model are output.

[0013] A further aspect of the present invention includes a coordinate calculation module, which performs the following operations: Calculate the distortion contribution value of each sensing node, and mark the nodes whose distortion contribution value exceeds the preset distortion threshold as polarization electrode channel combinations; Separate the row channel sequence number and column channel sequence number from the polarization electrode channel assembly; Call the capacitor matrix physical wiring mapping table to query the physical coordinates corresponding to the row channel sequence number and column channel sequence number; Calculate the center point of the physical coordinates to solve for the two-dimensional spatial coordinates.

[0014] A further aspect of the present invention includes an anomaly report generation module, which performs the following operations: Call the material type coding mapping table to convert the actual material type into a foreign object type identification code and touch shielding security level instruction; Convert two-dimensional spatial coordinates into binary data blocks; The foreign object type identifier, binary data block, and touch shielding security level instruction are assembled according to a predefined structured data frame format to generate a structured anomaly report.

[0015] In a further embodiment of the present invention, after the step of generating a structured anomaly report, the method further includes: Push structured anomaly reports to the main processor; and The main processor parses the structured anomaly report and, based on the two-dimensional spatial coordinates and the touch shielding security level instructions, drives the touch screen control system to perform adaptive anti-mistouch shielding on the screen area where the material type is located.

[0016] In summary, the present invention has the following beneficial technical effects: 1. This invention utilizes a frequency sweep excitation signal generation module to generate a continuous sequence covering a predetermined bandwidth, and constructs a complex impedance characteristic matrix based on this sequence, thereby obtaining broadband spectral response data of the local equivalent impedance network on the surface of a capacitive touchscreen. Since different materials possess inherent frequency-varying dielectric constants and conductivity characteristics, they exhibit unique frequency response trajectories in the complex plane. This provides richer feature dimensions compared to traditional single-frequency or discrete multi-frequency detection methods, which is beneficial for improving the system's ability to distinguish between materials with similar dielectric properties.

[0017] 2. This invention achieves effective separation of foreign object features from global environmental drift by calculating local geometric mapping features such as the radius of curvature of the resonant peaks and valleys and the tilt angles of the major and minor semi-axis of the complex impedance trajectory on the complex plane, combined with first-order difference operations. Since changes in environmental temperature and humidity typically cause overall linear translation or proportional scaling of the sensing node, the geometric features extracted by this invention focus on the morphological properties of the trajectory itself, which strongly suppresses such global fluctuations and helps maintain the reliability of detection results under different working environments.

[0018] 3. This invention, by tracing back the polarization electrode channel combinations that contribute to the distortion threshold and combining them with a preset capacitance matrix physical wiring mapping table, can associate the identified material type with the screen's physical coordinates. This mechanism realizes the conversion from logical channel numbers to two-dimensional spatial coordinates, providing structured positional data support for subsequent adaptive anti-mistouch shielding. This facilitates the system to perform precise touch function adjustments for specific contaminated areas, reducing the scope of false shielding. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention.

[0020] Figure 1 This is a schematic diagram of the framework in the embodiments of this application.

[0021] Figure 2 This is a flowchart illustrating an embodiment of this application. Detailed Implementation

[0022] The following is in conjunction with the appendix Figures 1-2 A preferred description of the present invention is provided below.

[0023] See attached document Figures 1-2 This invention proposes a foreign object detection and positioning system for touch screens based on capacitance matrix spectrum analysis, comprising the following modules: The frequency sweep excitation signal generation module generates a continuous frequency sweep excitation signal sequence covering a predetermined bandwidth according to the environmental monitoring instructions, and synchronously outputs the corresponding reference clock reference. The electrode driving and network forming module is used to perform time-sequential polling driving of the transmitting electrodes of the capacitive touch screen using a continuous frequency sweep excitation signal sequence, so as to establish a local equivalent impedance network that reflects the physical state of the surface of the capacitive touch screen. The phase-sensitive demodulation module is used to capture the analog signal modulated by the local equivalent impedance network and perform phase-sensitive demodulation on the analog signal based on the reference clock reference to extract the in-phase component signal and the quadrature component signal. The complex impedance matrix construction module calculates complex plane coordinate data based on in-phase and quadrature component signals, and combines the complex plane coordinate data to construct a complex impedance characteristic matrix. The geometric feature extraction module is used to process the complex impedance feature matrix to filter out global environmental changes and extract the local geometric mapping features of corresponding abrupt regions; The material classification module is used to compare local geometric mapping features with a pre-calibrated database of foreign object features to determine the type of substance in the mutation region. The coordinate calculation module is used to trace back the abnormal channel identifiers that contribute to the local geometric mapping features, and to use hardware array projection to calculate the two-dimensional spatial coordinates of the material type. The anomaly report generation module is used to stitch together the type of substance and two-dimensional spatial coordinates to generate a structured anomaly report that includes both qualitative and locational composite attributes.

[0024] In one embodiment of the present invention, the frequency sweep excitation signal generation module is used to perform the following steps: The system parses environmental monitoring commands to obtain a preset scanning parameter configuration file; adjusts the bias voltage of the voltage-controlled oscillator unit according to the preset scanning parameter configuration file to drive the voltage-controlled oscillator unit to output a continuous frequency sweep excitation signal sequence; extracts the phase and frequency parameters of the continuous frequency sweep excitation signal sequence and inputs them into a high-precision phase-locked loop to generate a reference clock reference that is synchronized with the continuous frequency sweep excitation signal sequence.

[0025] Specifically, the system's main control unit issues environmental monitoring commands via a serial communication bus. These commands include a start identifier and a preset scan parameter configuration file. The commands are received and decoded by the low-power microcontroller attached to the signal generator module. After decoding, the low-power microcontroller sends a wake-up signal to the power management sub-circuit, disconnecting the power supply connection to the bypass circuit and restoring the standard power supply voltage of the main signal link. This causes the signal generator module to switch from bypass standby mode to full-power operating mode. The low-power bypass standby mode refers to the signal generator module's non-core circuits being powered off, maintaining only the minimum power consumption state of the low-power microcontroller and the command receiving interface. The standard power supply voltage restored after wake-up is typically 3.3V or 5V.

[0026] Subsequently, the signal generator module calls its internally stored preset scan parameter configuration file. This configuration file defines the sweep start frequency, sweep end frequency, frequency step increment, single-point dwell time, and waveform modulation method. It should be noted that the sweep start and sweep end frequencies in the preset scan parameter configuration file define the complete excitation bandwidth range, typically set from 10 kHz to 1 MHz. This range covers the capacitive coupling frequency band common in capacitive touchscreen electrode networks and the effective frequency band where typical attachments may cause dielectric disturbances. The frequency step increment determines the precision of the sweep trajectory. Its setting is based on capturing sufficient feature points for shape analysis in subsequent complex impedance trajectory construction without significantly extending the scan time. Typical frequency step increment values ​​are set based on the Nyquist sampling principle and the system's real-time processing capabilities, typically ranging from 50 Hz to 500 Hz.

[0027] The dwell time at a single point is the duration for which the voltage-controlled oscillator (VCO) outputs stably at each discrete frequency point. Its setting should match the signal acquisition period of the subsequent receiving electrodes, typically ranging from 10 ms to 100 ms, to ensure that a sufficient number of stable signals for phase-sensitive demodulation can be acquired at each frequency point. The waveform modulation method is defaulted to sinusoidal modulation to ensure the spectral purity of the signal for subsequent analysis. The control logic unit generates a digital control sequence based on the configuration file. This sequence is input to the tuning voltage input of the VCO via a parallel interface to precisely adjust the bias voltage of the VCO. In this embodiment, the start voltage and stop voltage correspond to the tuning curve of the VCO. Their specific values ​​depend on the voltage tuning sensitivity coefficient K of the selected VCO model. For example, assuming a typical wide-tuning-range VCO chip used in the communication field is employed, its K coefficient is 20 MHz / V, covering a sweep frequency range of 10 kHz to 1 MHz. The calculated start voltage and stop voltage ranges are 0.1V to 0.5V and 4.5V to 5V, respectively.

[0028] The voltage step value is also calculated based on the frequency step increment and the K coefficient. The voltage value gradually increases from the starting voltage at the corresponding starting frequency to the ending voltage at the corresponding ending frequency according to a preset gradient. Each voltage step value is strictly matched with the frequency step increment, thereby driving the voltage-controlled oscillator (VCO) to output a continuous sweeping sinusoidal excitation signal sequence whose frequency increases monotonically from the starting point to the ending point according to a preset step size. Simultaneously, the phase synchronization circuit samples the original frequency signal output by the oscillator core of the VCO in real time and extracts the initial phase angle of its oscillation period as a reference value. The phase synchronization circuit inputs this reference value to a high-precision phase-locked loop (PLL). The high-precision PLL dynamically adjusts the frequency division coefficient and phase calibration circuit of its internal quartz oscillator according to the reference value and the real-time frequency parameters output by the VCO, outputting a digital reference clock signal that maintains phase alignment with the continuous sweeping excitation signal sequence at each frequency switching point and strictly tracks the frequency throughout the entire sweeping cycle, thus generating a reference clock reference.

[0029] It should be understood that high-precision phase-locked loops (PLLs) employ digital PLL technology. Their internal quartz oscillator reference frequency is typically 10 MHz or higher. A reference clock frequency is generated by dynamically adjusting the frequency division factor N, which is equal to the real-time output frequency of the voltage-controlled oscillator (VCO) multiplied by a fixed scaling factor M. The scaling factor M is usually a fixed integer, such as 10 or 100, to achieve strict time alignment. Phase alignment means that at each sweep step switching instant, the clock edge of the reference clock signal output by the PLL is consistent with a specific phase point of the sinusoidal excitation signal output by the VCO.

[0030] For example, suppose the system's main control unit issues an environmental monitoring command. Its preset scan parameter configuration file sets the sweep start frequency to 1 kHz, the sweep end frequency to 100 kHz, the frequency step increment to 500 Hz, the single-point dwell time to 50 ms, and the waveform modulation method to a sine wave. The low-power microcontroller decodes the command and wakes up the signal generator module. The signal generator module generates a digital control sequence according to the configuration file. Assuming the tuning sensitivity K of the voltage-controlled oscillator unit is 100 kHz / V, the starting voltage corresponding to the starting frequency of 1 kHz is 0.01V, and the ending voltage corresponding to the ending frequency of 100 kHz is 1V. To achieve a step increment of 500 Hz, the required voltage step value is 0.005V. The control sequence drives the tuning voltage of the voltage-controlled oscillator unit from 0.01V, increasing by 0.005V each time, reaching 1V after 200 steps. The output frequency starts from 1 kHz, increases by 500 Hz each time, and ends at 100 kHz, forming a continuous sweep excitation signal sequence. When the voltage-controlled oscillator unit outputs a signal at the first frequency point of 1 kHz, the phase synchronization circuit samples its initial phase, assuming that the measured initial phase angle is 30 degrees, and uses it as a reference.

[0031] Meanwhile, the high-precision phase-locked loop (PLL) uses a quartz oscillator with a reference frequency of 10 MHz and a scaling factor M of 100, meaning the reference clock frequency should be the excitation signal frequency multiplied by 100. For an excitation frequency of 1 kHz, the PLL needs to generate a 100 kHz reference clock. Its division factor N needs to be adjusted so that 10 MHz divided by 100 kHz equals 100. The PLL's phase calibration circuit adjusts the edge alignment of the output clock based on an initial phase reference of 30 degrees. When the excitation signal frequency steps to the next point of 1.5 kHz, the PLL adjusts N in real time to approximately 66.67 (10 MHz divided by 150 kHz) while maintaining phase alignment, and so on, synchronously generating the reference clock reference. The entire process generates an excitation signal sequence covering 200 discrete frequency points from 1 kHz to 100 kHz, along with a strictly aligned reference clock sequence.

[0032] In one embodiment of the present invention, the electrode driving and network forming module is configured to perform the following steps: A scan timing generator is driven by a reference clock reference to generate an array scan timing signal. The array scan timing signal is used to control an analog multiplexed switch array to sequentially turn on each row of the capacitive touch screen's transmitting electrodes. A continuous sweep frequency excitation signal sequence is injected into the currently turned-on transmitting electrode. The dielectric coupling effect of the screen surface attachments on the edge electric field is used to establish a local equivalent impedance network at the spatial intersection of the transmitting and receiving electrodes.

[0033] Specifically, the system receives a reference clock as a global synchronization clock source, driving the frequency divider and counter inside the scan timing generator to produce an array scan timing signal that is strictly synchronized with the reference clock and repeats periodically. The array scan timing signal consists of a series of square wave pulses with fixed widths. The width of each pulse is equal to the dwell time of the continuous sweep excitation signal sequence at each frequency point. The intervals between pulses are sufficiently short to ensure switching efficiency, and the period of the entire pulse sequence is equal to the total time required to scan and cover all row emitter electrodes of a capacitive touchscreen. It should be noted that the pulse width of the array scan timing signal is strictly equal to the dwell time at each point to ensure that a complete excitation and signal acquisition cycle for a row of emitter electrodes of a capacitive touchscreen can be completed at each scan frequency point.

[0034] The total time required to scan all row electrodes of a capacitive touchscreen is calculated by multiplying the total number of rows by the pulse width. The pulse interval is set to ensure that the switching action of the multiplexer is electrically stable and does not introduce signal crosstalk. Typical interval times are based on the maximum switching times (tr and tf) of the selected analog switches, typically ranging from 1 μs to 10 μs. The scan timing generator synchronously transmits this pulse sequence via a parallel data bus to the control pins of an array of analog multiplexers integrated at the end of the screen driver circuit. The model of the analog multiplexer array must meet the low-loss requirements of high-frequency signal transmission, for example, by using an integrated switch chip based on CMOS technology. The analog multiplexer array internally includes switch control decoding logic. Upon receiving the valid edge of the Nth pulse in the array scan timing signal, this decoding logic decodes and activates the corresponding switch channel for the Nth row emitter electrode, while simultaneously closing all other non-target channels, thereby sequentially and uniquely turning on each row emitter electrode of the capacitive touchscreen. When the switch channel corresponding to a certain row of emitter electrodes is activated and turned on, the current frequency point signal in the continuous sweep excitation signal sequence is boosted and shaped by the signal drive amplifier circuit with gain buffer and impedance matching functions, and then injected into the currently turned emitter electrode channel through the physical wiring of the turned-on switch channel.

[0035] It should be understood that the gain of the signal-driven amplifier circuit is set to a fixed value, designed to boost the voltage amplitude of the continuous sweep excitation signal sequence to a level sufficient to drive the electrode network and overcome line losses; typical gain values ​​are 2 to 5 times. The contacting emitter electrode is equivalent to an excitation source node. Driven by a high-frequency sinusoidal excitation signal, the parasitic capacitance formed by the electrode metal traces and the touch glass cover medium around it excites an edge electric field that penetrates the touch glass cover and diffuses across its surface. This edge electric field is a non-uniform electric field distribution generated by the high-frequency AC signal at the electrode tip and the boundary of the cover medium due to the concentration of electric field lines; its intensity is related to the frequency and amplitude of the excitation signal. When there are deposits on the screen surface, these deposits, due to their different dielectric constants and conductivity characteristics, will exert dielectric coupling interference on the intensity and distribution of the edge electric field. This interference alters the spatial characteristics of the electric field coupling. The dielectric coupling interference refers to the fact that the dielectric constant εr and conductivity σ of the attached material differ from those of air. When placed in an electric field, it alters the electric field path and energy distribution. This change can be equivalent to adding an additional complex impedance in parallel between the original electrodes. This altered electric field characteristic ultimately induces a complex electrical effect in the spatial intersection region between the currently driving transmitting electrode and the adjacent receiving electrode in receiving mode. This effect can be modeled in circuit analysis as a two-port network composed of parallel capacitive and resistive components. This network is the local equivalent impedance network carrying the dielectric constant and conductivity characteristics of the attached material. This local equivalent impedance network is an abstract circuit model used to characterize the complex impedance characteristics of the electrical path between a specific transmitting electrode and a specific receiving electrode under specific frequency excitation. This impedance Z is a complex number, including the reactance corresponding to the real resistive component R and the imaginary capacitive component C.

[0036] For example, continuing the previous example, the continuous sweep excitation signal sequence includes 200 frequency points, each with a dwell time of 50 ms. Assume the capacitive touchscreen has 16 rows of transmitting electrodes. A scan timing generator is driven by a reference clock as the synchronization source. The reference clock corresponds to a frequency of 100 kHz at the first frequency point of 1 kHz. The scan timing generator uses this 100 kHz clock to generate a pulse sequence via an internal counter: each pulse is 50 ms wide, the pulse interval is 5 μs, and the entire sequence consists of 16 pulses, corresponding to the 16 rows of electrodes, with a sequence period of 16 times 50 ms, or 800 ms. During the sweep at each frequency point, this 800 ms periodic pulse sequence is repeated to ensure that all 16 rows of electrodes are scanned sequentially at each frequency point.

[0037] At the start of the first 1 kHz scan cycle, the first pulse edge of the array scan timing signal reaches the analog multiplexed switch array. After decoding by the switch control decoding logic, the switch channel controlling the first row of transmitting electrodes is activated, turning on the first row of electrodes. At this time, a 1 kHz sinusoidal signal with a set amplitude from the continuous frequency sweep excitation signal sequence is amplified by a signal drive amplifier with a gain of 3 and injected into the already turned-on first row of transmitting electrode channels. This electrode is excited, generating an edge electric field around it. Assume that at this time, there is a water droplet attached to the screen surface at the junction of the first row of electrodes and the adjacent receiving electrode. The dielectric constant εr of this water droplet is approximately 80, and the conductivity σ is approximately 0.01 S / m, which couples and interferes with the edge electric field. This interference changes the electrical characteristics between the first row of transmitting electrodes and the specific receiving electrode, establishing a local equivalent impedance network.

[0038] After the first 50 ms pulse ends, the timing pulse switches, and the second pulse turns on the second row of electrodes, injecting the same 1 kHz excitation signal to establish a local equivalent impedance network for the second row of electrodes. This process continues until the 16th row of electrodes is scanned. After completing one round of polling for the 16th row of electrodes, the process moves to the next frequency point of 1.5 kHz and repeats the above 800 ms cycle scanning timing to perform a new round of polling for the 16 rows of electrodes, sequentially establishing the local equivalent impedance network for each electrode at each frequency point. This process continues throughout the entire 200 frequency point sweep.

[0039] In one embodiment of the present invention, the phase-sensitive demodulation module is used to perform the following steps: The analog signal and the reference clock reference are input to an analog multiplier for multiplication to generate a mixed signal; the mixed signal is input to a filter to remove high-frequency harmonic components and obtain a low-frequency DC component; the low-frequency DC component is separated to extract the in-phase component signal representing the real part of the complex impedance and the quadrature component signal representing the imaginary part of the complex impedance.

[0040] Specifically, after stimulating a row of transmitting electrodes and establishing a local equivalent impedance network, the signal acquisition process for each column of receiving electrode channels corresponding to the spatial position of that row of electrodes is immediately initiated. The control logic sends an acquisition command to the analog-to-digital converter (ADC) array, which includes multiple independent ADC channels, each physically connected to a receiving electrode. The command activates the specific ADC channel connected to the target receiving electrode, and the sample-and-hold circuit within that channel begins operation, capturing the analog voltage signal modulated by the local equivalent impedance network and fed in via the receiving electrode in real time at a fixed sampling rate.

[0041] It should be noted that the sampling rate of the analog-to-digital conversion channel must be set to be higher than twice the highest frequency in the continuous sweep excitation signal sequence to satisfy the Nyquist sampling theorem and ensure distortion-free signal capture. A typical sampling rate is set based on the sweep termination frequency; for example, if the sweep termination frequency is 100 kHz, the sampling rate should be set to at least 200 kHz. The analog voltage signal is the voltage response induced on the receiving electrodes after the continuous sweep excitation signal sequence crosses the screen medium and is disturbed by the local equivalent impedance network. The changes in its amplitude and phase carry information about the complex impedance changes of the local equivalent impedance network.

[0042] Subsequently, a reference clock reference is input as a synchronization signal source to the first input port of the analog multiplier integrated circuit, while an analog voltage signal read from the analog-to-digital conversion channel buffer is input to the second input port of the analog multiplier. The analog multiplier integrated circuit is the core component for phase-sensitive demodulation, typically a four-quadrant analog multiplier chip. Its operating principle is based on the multiplication characteristics of analog circuits to multiply the instantaneous values ​​of two input signals. The analog multiplier performs the multiplication operation of the two input signals, generating a mixed signal including the sum and difference frequency components of the original signal frequency and the reference clock frequency. This mixed signal is then fed into a low-pass filter circuit with a configurable cutoff frequency. It should be understood that the cutoff frequency of the low-pass filter is a critical configuration parameter, set to retain low-frequency information related to impedance changes while filtering out high-frequency components. After multiplication, the low-frequency difference frequency component in the sum and difference frequency components includes the amplitude and phase information of impedance change, while the high-frequency sum frequency component is noise. Therefore, the cutoff frequency is usually set to one-tenth of the frequency value of the current scanning frequency point. For example, for a scanning frequency of 1 kHz, the cutoff frequency can be set to 100 Hz.

[0043] The low-pass filter is configured to allow only low-frequency components with frequencies significantly lower than the current frequency point of the continuously swept excitation signal sequence to pass through, thereby effectively filtering out high-frequency harmonic noise generated by multiplication operations and other high-frequency interference that may be carried in the signal. After filtering, only slowly changing DC or near-DC low-frequency components remain in the output signal. This filtered output is then processed by a DC component extraction circuit, which typically includes a gain buffer stage composed of high-precision operational amplifiers and a simple RC network for isolating residual AC components. The time constant of the RC network must be much larger than the lowest change period of the signal to ensure stable extraction of the DC level.

[0044] Ultimately, the signal output from the DC component extraction circuit is separated into two independent voltage values. The first voltage value represents the in-phase component with zero phase difference between the analog voltage signal and the reference clock reference. The magnitude of this component directly reflects the real part of the complex impedance of the local equivalent impedance network at the current scanning frequency. The second voltage value represents the quadrature component with a 90-degree phase difference. The magnitude of this component reflects the imaginary part of the complex impedance of the local equivalent impedance network. Subsequently, by vector synthesis of the in-phase and quadrature components, the impedance amplitude and phase deflection angle can be further derived. The real and imaginary parts together constitute a complete description of the complex impedance. More specifically, the voltage value of the in-phase component signal corresponds to the component in the demodulated signal that is in phase with the reference clock, and its magnitude is proportional to the impedance amplitude of the local equivalent impedance network. The voltage value of the quadrature component signal corresponds to the component in the demodulated signal that is orthogonal to the reference clock, and its magnitude is proportional to the sine of the phase deflection angle of the local equivalent impedance network. The impedance amplitude and phase deflection angle together constitute a complete description of the complex impedance.

[0045] For example, when scanning the first row of transmitting electrodes with an excitation frequency of 1 kHz at the first point, a local equivalent impedance network is established for the region where this row of electrodes intersects with all receiving electrodes. Assume the current focus is on the first column of receiving electrode channels. The control logic activates the analog-to-digital conversion channel corresponding to the first column of receiving electrodes. This channel captures an analog voltage signal at a sampling rate set to 200 kHz. Assume that due to the influence of water droplets adhering to the screen surface, the captured signal is a sine wave with an amplitude of 50 mV and a phase offset of 45 degrees relative to the original 1 kHz excitation signal.

[0046] Simultaneously, a reference clock signal with a frequency of 100 kHz, strictly synchronized with the 1 kHz excitation signal, is input to an analog multiplier. The analog multiplier multiplies the 1 kHz signal (50 mV with a 45-degree phase shift) with the 100 kHz reference clock signal. The multiplication operation produces two main frequency components: a sum frequency component of 101 kHz and a difference frequency component of 99 kHz. This mixed signal is fed into a low-pass filter with a cutoff frequency set to 100 Hz. The filter significantly attenuates the 101 kHz and 99 kHz components, allowing only extremely low-frequency components below 100 Hz to pass. After filtering, the output signal mainly consists of DC levels. The DC component extraction circuit further processes this signal and separates it into two voltage values. According to the phase-sensitive demodulation principle, the in-phase voltage value corresponds to the real part of the complex impedance, and the quadrature voltage value corresponds to the imaginary part of the complex impedance.

[0047] Assuming that after demodulation, the in-phase component signal voltage is 30mV and the quadrature component signal voltage is 10mV, the combination of these 30mV and 10mV values ​​represents the real and imaginary parts of the complex impedance of the local equivalent impedance network established by the water droplet attachment at the intersection of the first row of transmitting electrodes and the first column of receiving electrodes at a 1 kHz scanning frequency. This in-phase and quadrature component data is stored in a buffer. This process is performed in parallel or sequentially for each column of receiving electrodes to obtain complete impedance information at the intersection of the current row of electrodes and all columns of electrodes at the current frequency. Subsequently, when the scanning timing switches to the next row of electrodes or the next frequency, the above process of activating analog-to-digital conversion, multiplicative demodulation, and component extraction is repeated.

[0048] In one embodiment of the present invention, the complex impedance matrix construction module is used to perform the following steps: Using the values ​​of the in-phase component signal as the real axis coordinates and the values ​​of the quadrature component signal as the imaginary axis coordinates, the complex impedance points of a single sensing node at a specific frequency are calculated. Multiple complex impedance points measured at all scanning frequency points of the same sensing node are connected in ascending order of frequency to generate a continuous impedance trajectory array. All sensing nodes on the screen are traversed, and all continuous impedance trajectory arrays are combined to generate a complex impedance characteristic matrix.

[0049] Specifically, after completing phase-sensitive demodulation of all sensing nodes and extracting the in-phase and quadrature component signals at each frequency point, each pair of component data is temporarily stored in a two-dimensional data buffer organized by node index and frequency index. It should be noted that the sensing reference node here refers to the unique sensing unit on the capacitive touchscreen formed by the intersection of one transmitting electrode and one receiving electrode, its physical location defined by both row and column numbers. Subsequently, the data processing unit sequentially reads the value of the in-phase component signal decoupled from each sensing node at each specific scanning frequency point from the buffer and directly assigns it as the real part Re of the complex coordinate system; simultaneously, it reads the value of the corresponding quadrature component signal and assigns it as the imaginary part Im of the complex coordinate system. In this embodiment, the values ​​of the real part Re and the imaginary part Im of the complex impedance point Z are directly derived from the voltage values ​​of the aforementioned extracted in-phase and quadrature component signals. Their value range is determined by the quantization range and signal amplitude of the analog-to-digital converter. Typically, normalization processing is required, and the normalization coefficient is usually set based on the system's maximum output voltage or full-scale voltage.

[0050] Based on this, the complex impedance point of the sensing node at that specific frequency can be calculated. Its mathematical definition is ,in This represents the imaginary unit. For each sensing node, the two-dimensional data buffer is accessed sequentially according to the scanning frequency index from low to high to obtain multiple complex impedance points calculated for that node at all scanning frequency points. Then, a trajectory smoothing connection algorithm is executed. This algorithm reads the coordinate values ​​of the complex impedance points corresponding to two adjacent frequency points according to the index order between adjacent frequency points, and performs linear interpolation calculation between these two coordinate points according to a preset interpolation density to generate a series of intermediate virtual complex impedance points, thereby converting the discrete sequence of complex impedance points into a smooth curve with continuously varying frequency on the complex plane.

[0051] It should be understood that the trajectory smoothing connection algorithm aims to eliminate data point discontinuities caused by discrete frequency sampling, making the trajectory of impedance changing with frequency more continuous and facilitating subsequent feature analysis. Commonly used interpolation algorithms include linear interpolation or cubic spline interpolation. The interpolation density parameter determines the number of intermediate points generated, which is usually set to insert within each original frequency interval. One point, The value range is typically from 3 to 10, depending on the trajectory smoothness requirements of subsequent analysis. The specific process of linear interpolation is as follows: for two adjacent frequency points... and Corresponding complex impedance point and Insert in between The point, the first Complex impedance value at each interpolation point It can be calculated using the following formula:

[0052] in, This represents the first value calculated between two original measurement points. Interpolation complex impedance points. Indicates the starting frequency point The original complex impedance value measured at the location. Indicates the next adjacent frequency point The original complex impedance value measured at the location. This represents the interpolation density, which is the total number of equal intervals divided between two original frequency points. This represents the index of the currently calculated interpolation point in the linear sequence, with values ​​ranging from 1 to... .

[0053] All points on this smooth curve, including the original measurement points and interpolation points, are stored in a one-dimensional array in ascending frequency order. This array is the continuous impedance trajectory array reflecting the frequency variation trend of the node's impedance. All predefined sensing reference nodes on the capacitive touchscreen are traversed. Each node corresponds to a physical location identifier uniquely determined by its row transmitting electrode number and column receiving electrode number. For each node, the system repeats the process of calculating complex impedance points and constructing the continuous impedance trajectory array. Finally, the continuous impedance trajectory arrays of all nodes are combined and mapped: using the node index as the first dimension, the frequency point index (including the equivalent frequency index of the interpolation points) as the second dimension, and the index corresponding to each point (including the real part) as the third dimension. and the virtual part The complex impedance values ​​of the two components form the third dimension of the data, constructing a three-dimensional data structure. This data structure is the complex impedance characteristic matrix characterizing the global electromagnetic dynamic response. The complex impedance characteristic matrix is ​​a three-dimensional data structure, where the size of its first dimension is equal to the total number of inductive reference nodes. The size of the second dimension is equal to the total number of scan frequency points. Multiply by interpolation density The third dimension includes two channels that store the real and imaginary parts, respectively. This matrix fully records the complex impedance response state of the screen at all node positions and at all scanning frequencies across the entire screen.

[0054] For example, continuing from the previous example, suppose the system has acquired the in-phase component signal value of the first row and first column sensing reference node at the first scan frequency point of 1 kHz as 30mV and the quadrature component signal value as 10mV. The data processing unit reads this data and assigns 30mV to the real part. A value of 10mV is assigned to the imaginary part Im, and the complex impedance of the node at a frequency of 1kHz is calculated. Its value is The unit is mV. Assume the system continues to scan subsequent frequency points and acquires similar data for that node at 1.5 kHz, 2 kHz... up to 100 kHz. For example, at 1.5 kHz, the in-phase component is measured to be 35 mV and the quadrature component 15 mV, then the complex impedance point... for The system visits these discrete points in ascending order of frequency: , ..., until the last frequency data point. Then, the system performs a smooth connection. The interpolation density M is set to 5. For the corresponding frequency... of and corresponding frequency of The system performs linear interpolation between these two adjacent points to calculate the frequency interval. Interpolation step size .for Interpolation frequency Interpolated complex impedance Similarly, calculate the interpolation points for k from 2 to 4.

[0055] Finally, for this node, a continuous impedance trajectory array is generated, including a sequence of complex impedance values ​​from all original measurement points from 1 kHz to 100 kHz, as well as interpolated intermediate points. Next, all inductive reference nodes are traversed; assuming the screen has 16 rows and 20 columns, or 320 nodes, the same calculation and interpolation process is performed on each node to generate its own continuous impedance trajectory array. Finally, the trajectory arrays of all 320 nodes are combined. Using the node index as the first dimension (range 1 to 320) and the frequency index as the second dimension, with a total number of points (e.g., 200 original points multiplied by an interpolation density of 5), approximately 1000 equivalent frequency indices are constructed. Each index position stores a complex impedance value including both real and imaginary parts, thus constructing a complex impedance characteristic matrix of size 320 × 1000 × 2, which characterizes the global electromagnetic dynamic response.

[0056] In one embodiment of the present invention, the geometric feature extraction module is configured to perform the following steps: The squared Euclidean distance between the complex impedance characteristic matrix and the preset cleanroom reference matrix at each frequency point is calculated to form an Euclidean distance evolution sequence. A first-order difference operation is performed on the Euclidean distance evolution sequence to eliminate the overall linear translation of the trajectory caused by changes in environmental temperature and humidity. Based on the processed trajectory data, the resonant peak and valley points are located, and the radius of curvature and the tilt angle of the major and minor semi-axis at the resonant peak and valley points are calculated to form local geometric mapping features.

[0057] Specifically, the data processing unit loads a preset cleanroom reference matrix from non-volatile memory. This matrix has the same structure as the complex impedance characteristic matrix, possessing the same dimensions for the total number of nodes, the total number of frequency points, and the complex impedance values. The preset cleanroom reference matrix is ​​obtained by performing the same complete frequency sweep detection process—from the frequency sweep excitation signal generation module to the complex impedance matrix construction module—multiple times under the standard state of a clean, uncontaminated capacitive touchscreen. The arithmetic mean of all collected complex impedance characteristic matrices is then stored. First, all sensing reference node indices are traversed. For each node index, the squared Euclidean distance is calculated point-by-point along the frequency index dimension.

[0058] Specifically, for node index i and frequency index j, the position in the complex impedance characteristic matrix is ​​read. Complex impedance value at Simultaneously read the same position in the preset cleanroom reference matrix. Reference complex impedance value at Calculate the square of the modulus of the difference between these two complex numbers, and use it as the square of the instantaneous Euclidean distance at that frequency point. The calculation formula is as follows:

[0059] In the formula, Indicates the first The first sensing node, the first The instantaneous Euclidean distance squared value at each scanning frequency point is used in this scheme to simplify the calculation complexity and retain the monotonicity of the deviation characteristics. This quantifies the degree of difference between the measured impedance and the reference impedance at that point. This represents the index of the inductive reference node in the capacitance matrix. This indicates the index of a frequency point in the scanned frequency sequence. Indicates at node ,frequency The complex impedance value in the complex impedance characteristic matrix obtained from actual measurement. Indicates at node ,frequency The reference complex impedance value is read from the preset cleanroom reference matrix.

[0060] After calculating all frequency index points for a node, then all points... The values ​​are summed to obtain the total Euclidean distance value of the node. The calculation formula is as follows:

[0061] In the formula, Indicates the first The total Euclidean distance of a sensing node across all scanning frequency points is a comprehensive measure of the node's overall deviation from the baseline state. Indicates the index of the sensing reference node. That is, the squared instantaneous Euclidean distance as defined above. This indicates the total number of scan frequency points.

[0062] Simultaneously record the values ​​at each frequency point. They are arranged in ascending order of frequency to form the Euclidean distance evolution sequence of that node. Essentially, it is a vector that records the nodes. The degree of deviation of its impedance response from the reference state at different frequencies. Then, the evolution sequence of all nodes is traversed. Mutation pattern identification is performed for each sequence. Calculate its arithmetic mean over the entire frequency range. and standard deviation Set a mutation detection threshold. The threshold is equal to Plus times The specific formula is as follows:

[0063] In the formula, Represented as the first A mutation threshold is set for each node to determine whether there are significant anomalies in the Euclidean distance evolution sequence of that node. This threshold is set based on statistical principles; slow environmental drift typically leads to mutations in all nodes. The sequence revolves around its mean Fluctuates within a relatively small range, standard deviation The smaller the frequency, the more significant the deviation at certain frequency points due to the presence of abrupt changes in the region containing the attached material. Indicates the first Euclidean distance evolution sequence of nodes The arithmetic mean, By all Arranged in order of frequency. Indicates the first Euclidean distance evolution sequence of nodes The standard deviation. This represents the preset multiplier coefficient used to adjust the sensitivity of the threshold. Typically, settings are based on experimental data, such as by analyzing a large number of normal samples. The value is 2 to 5 to ensure that normal fluctuations can be effectively distinguished from abnormal mutations.

[0064] Traversal sequence Each of them Value, statistics exceeding it The proportion of frequency points to the total number of frequency points .like Greater than the preset ratio threshold If the node trajectory is determined to be a mutated node trajectory that deviates from the uniform environment drift pattern, its node identifier is added to the mutated node list. (Preset proportional threshold) To determine whether a node has deviated overall, an empirical value is set, such as 20% to 50%, indicating that a node is considered mutated only if it exhibits significant anomalies at a sufficient number of frequency points. Based on the list of mutated nodes, the complete continuous impedance trajectory arrays of the corresponding nodes are extracted from the complex impedance feature matrix, forming a mutated node trajectory dataset. For each extracted mutated node trajectory data, a first-order difference operation is performed to remove noise. The difference operation targets the Euclidean distance evolution sequence corresponding to that trajectory. For the sequence Subtract adjacent elements from the given sequence to generate a new difference sequence. The formula for calculating the first-order difference sequence is:

[0065] In the formula, Indicates the first The first-order difference sequence of the nth node This operation is used to highlight the local slope of distance changes in order to reduce the overall linear drift caused by environmental changes. Indicates the first The th node in the Euclidean distance evolution sequence Each element, namely . This represents the next element in the Euclidean distance evolution sequence, i.e. . Indices representing frequency points. This operation aims to eliminate... The sequence may contain trends of overall linear translation or proportional scaling, as these trends may be caused by slow changes in ambient temperature and humidity. Difference can highlight local mutation features.

[0066] differential sequence The local slope, reflecting changes in distance, can mitigate the effects of slow changes in ambient temperature and humidity. The sequence exhibits an overall linear shift or proportional scaling trend. Next, for this variation node, the original continuous impedance trajectory array is extracted from the complex impedance characteristic matrix; that is, the sequence of complex impedance values ​​varying with frequency. Draw on the complex plane The impedance amplitude sequence is analyzed, and the resonant peaks and valleys are located. Resonant peaks and valleys refer to the frequency points where the impedance amplitude exhibits local maxima and minima on the complex impedance trajectory. Their location is achieved by applying a sliding window to the impedance amplitude sequence and comparing the values ​​of the center point within the window with those of its neighboring points. The location method involves analyzing the impedance amplitude sequence... Perform local extremum detection to find peaks that satisfy the local maximum condition and valleys that satisfy the local minimum condition.

[0067] For each located peak or valley point, a trajectory point within a small neighborhood window is selected. The impedance amplitude and frequency data within this window are mapped to the dimensionless feature space [0,1] to form dimensionless trajectory data. A quadratic polynomial is then used to fit the dimensionless trajectory curve within this window. Based on the polynomial coefficients obtained from the fitting, the radius of curvature at the peak or valley point is calculated. The calculation is used to quantify the curvature of the trajectory near the extreme points, for a function The plane curve described, which lies at the point radius of curvature at The calculation formula is:

[0068] In the formula, The radius of curvature represents the degree of curvature of the curve at a given point. In this paper, this formula is used to calculate the curvature of the complex impedance trajectory at the resonant peak and valley points. This indicates the independent variable, which in this context usually refers to frequency. This represents the function describing the curve. In this paper, it can be a function of the amplitude or phase of the impedance as a function of frequency, typically obtained by polynomial fitting of local trajectory points. Representation function right The first derivative. Representation function right The second derivative is obtained by fitting a quadratic polynomial and substituting it into the curvature formula. The size of the fitting window for the quadratic polynomial needs to be reasonably set according to the frequency step and the rate of change of impedance, for example, covering 5 to 10 adjacent frequency points.

[0069] At the same time, The overall shape of the sequence is considered as an approximate elliptical profile. An ellipse fitting algorithm is used to calculate the lengths of the major and minor semi-axes of the best-fit ellipse, as well as the tilt angle of the major axis relative to the real axis of the complex plane. The tilt angles of the major and minor semi-axes are obtained through the ellipse fitting algorithm, which fits the coordinate points of the complex impedance trajectory into an ellipse equation, and then solves for the length of the major axis *a*, the length of the minor axis *b*, and the angle θ between the major axis and the x-axis. Ellipse fitting typically uses the least squares method to fit the general quadratic equation of the ellipse.

[0070] Finally, the calculated array of curvature radii for all resonant peaks and valleys at the node, along with the semi-major axis length, semi-minor axis length, and tilt angle parameters obtained from ellipse fitting, are combined into a multi-dimensional feature vector. This feature vector represents the local geometric mapping feature independent of slow-drift environmental factors. This local geometric mapping feature vector includes the curvature radius array and ellipse parameters; these features are insensitive to slow environmental drift and primarily capture abrupt changes in local geometry caused by attachments.

[0071] For example, a preset cleanroom reference matrix is ​​loaded, which is also a 320×1000×2 three-dimensional matrix. Assume the system is currently analyzing the sensing reference node with index 1. The data for this node at all frequency indices is traversed. For frequency index 1, corresponding to 1 kHz, the data is read from the complex impedance characteristic matrix. for Read from the reference matrix for Calculate the squared value of the instantaneous Euclidean distance Calculate the value of this node across all 1000 frequency indices sequentially. Value. Assume that the total Euclidean distance is obtained by summing. Simultaneously, an Euclidean distance evolution sequence is formed. .calculate mean of the sequence Standard deviation Set the multiplier. Then the mutation determination threshold .

[0072] statistics More than 24 in the sequence Proportion of values Suppose that statistics show 400 points exceeding the threshold, representing 40% of the total 1000 points, which is greater than the preset threshold. For example, if the value is 30%, then node 1 is determined to be a mutated node and added to the list. Extract the continuous impedance trajectory array of node 1. For node 1 Perform first-order difference operations on sequences: for example , ,but The difference sequence is calculated sequentially. Next, processing is performed on the complex plane. Array. Detection impedance amplitude sequence Consider a local extremum, assuming a peak with an amplitude of 50mV is found at frequency index 200. Perform a quadratic polynomial fitting on five points near this peak. Assume the fitted function yields... Calculate the derivative at the peak point. and Substituting into the curvature formula, the radius of curvature is calculated. It is approximately 0.05.

[0073] At the same time, for the entire Perform an ellipse fitting on all 1000 points of the array, assuming the fitting yields the major semi-axis of the ellipse. short half-shaft Inclination angle Degrees. Finally, the peak curvature radius of 0.05, along with the curvature radii of other possible valley points, are combined with the ellipse parameters. , , The degrees are combined to form a feature vector [0.05,...,35,20,30], which serves as the local geometric mapping feature of node 1. This process is repeated for all other nodes in the mutated node list to generate a local geometric mapping feature vector for each mutated node.

[0074] In one embodiment of the present invention, the material classification module is used to perform the following steps: The nearest neighbor pattern recognition algorithm is used to calculate the error distance between the local geometric mapping features and the standard feature vectors in the foreign object feature database. The minimum value in the error distance is compared with a preset error tolerance threshold. When the minimum value is less than the preset error tolerance threshold, the standard feature vector corresponding to the minimum value is determined as the matching result, and the circuit distribution attributes corresponding to the matching result are retrieved. The circuit distribution attributes are analyzed to reconstruct the equivalent topological physical model, and the type of substance mapped by the equivalent topological physical model is output.

[0075] Specifically, it connects to a non-volatile storage medium via an internal data bus or network interface. This storage medium contains a pre-calibrated foreign matter feature database. It should be noted that this pre-calibrated database is established through a laboratory calibration process. The calibration process involves placing standard substance samples of known types and physical properties sequentially on a clean capacitive touchscreen surface. For each sample, a complete detection process, from the frequency sweep excitation signal generation module to the geometric feature extraction module, is performed to obtain the local geometric mapping feature vector corresponding to that substance. The experiment is repeated multiple times for each sample, and the arithmetic mean of its feature vectors is used as the final standard vector for that substance. Stored in the database. Simultaneously, based on the known electrical properties of the material and its interaction physical model with the electrodes, its equivalent circuit distribution properties are derived or fitted. And associated storage. The selection of representative materials covers the main types of foreign matter that may be encountered in the expected application scenarios, with discrete dielectric constant values ​​ranging from air close to 1 to water around 80.

[0076] The foreign matter characteristic database is a structured, multidimensional dataset. Its core data table records a series of representative material standard parameters with different discrete dielectric constant values. Each database record consists of three main parts: a unique material identifier... The standard local geometric mapping feature vector corresponding to this substance And a string describing the circuit distribution properties equivalent to the frequency response characteristics of the material. Standard local geometric mapping eigenvectors The dimension is exactly the same as the aforementioned output local geometric mapping feature vector, and its elements include an array of resonant peak and valley curvature radii and ellipse fitting parameters. Circuit distribution attribute description string. The equivalent circuit topology of the material's influence on the capacitor screen electrode network within a specific frequency range is defined, for example, "parallel RC model, R=100kΩ, C=5pF". Circuit distribution attribute description string. This is a simplified representation that transforms the physical properties of a material into circuit network parameters. Its format is a predefined text template, with parameters appended as key-value pairs. Parsing this string generates the corresponding circuit netlist in the program, which can be used for signal simulation or mechanism analysis.

[0077] During retrieval, the local geometric mapping feature vector is used. As the input for the query. The nearest neighbor pattern recognition algorithm is specifically the case where k=1 in the k-nearest neighbor algorithm, that is, calculating... With all in the database The Euclidean distance between them. Its calculation formula is:

[0078] in Represents the query vector With the database standard vectors The Euclidean distance between them. and Representing the first two vectors respectively Each characteristic element, such as the radius of curvature value or ellipse parameter. This refers to the total dimension of the feature vectors. The Euclidean distance is calculated by summing the squared differences across all dimensions and then taking the square root. The result is a scalar value used to measure the overall similarity between two vectors in the feature space; a smaller value indicates higher similarity. Since the elements in the feature vectors may differ in physical meaning and numerical dimensions (e.g., the radius of curvature is dimensionless, while the lengths of the major and minor axes are voltage-dimensional), it is necessary to consider the physical dimensions of the feature vectors before calculating the distance. and all Perform uniform normalization preprocessing, such as z-score standardization, so that the mean of each feature dimension is 0 and the standard deviation is 1, to eliminate the dominant influence of different units and numerical ranges on the distance metric.

[0079] For the m-th record in the database, its standard vector is Calculate the error distance Iterate through all M records in the database to obtain M items. The distance array of values. Then, the smallest value is found in this distance array. And record its corresponding substance identifier. In this embodiment, an error tolerance threshold can be set. If the smallest Still greater than If it is, then it is determined to be an unknown substance. The design is based on cross-validation testing of known substance samples, taking the minimum distance value among all incorrect matches as the safety boundary. Typical values ​​can be obtained through experimental statistics; for example, a normalized distance value of 0.5. The actual substance types are pre-stored, human-readable label information in the database records. The corresponding standard trajectory feature vector This refers to the matching result, specifically the corresponding circuit distribution attribute description string. Retrieved. Analysis. The string is used to extract the circuit component types, connection methods, and parameter values. Based on this, a corresponding equivalent circuit topology physical model is instantiated in the software. This model is used to interpret the source of the currently detected anomalous signal. Finally, the output is... The name of the actual substance type that is directly associated and stored in another field of the database, such as deionized water, silicone grease, or metal dust.

[0080] For example, the system has extracted the local geometric mapping feature vector of node 1. The range is [0.05,…,35,20,30], assumed to be normalized; the value here is the normalized value. The system is connected to a foreign matter characteristic database. Assume the database contains 3 records: record 1 corresponds to deionized water, Record 2 corresponds to the silicone grease. Record 3 corresponds to the metal shavings. .calculate Calculate the Euclidean distance to each record. Calculate the distance to record 1: Assume the summation result is... .

[0081] Similarly, calculation and Assume the results are 8.50 and 15.30 respectively. In the distance array [2.47, 8.50, 15.30], the minimum value is... The value is 2.47, corresponding to the substance identifier. It is deionized water. The standard trajectory feature vector for determining a matching hit is: Retrieve the circuit distribution attribute string associated with this record. Assuming the input is a "parallel RC model, R=150kΩ, C=3.5pF", this string is used to reconstruct the equivalent topological physical model in memory, consisting of a 150kΩ resistor and a 3.5pF capacitor connected in parallel. Finally, the output is the substance directly mapped by this model: deionized water.

[0082] In one embodiment of the present invention, the coordinate calculation module is used to perform the following steps: Calculate the distortion contribution value of each sensing node, and mark the nodes whose distortion contribution value exceeds the preset distortion threshold as polarization electrode channel combinations; extract the row channel sequence number and column channel sequence number from the polarization electrode channel combination; call the physical wiring mapping table of the capacitor matrix to query the physical coordinates corresponding to the row channel sequence number and column channel sequence number; calculate the center point of the physical coordinates to solve the two-dimensional spatial coordinates.

[0083] Specifically, after completing the determination of the actual material type and the reconstruction of the equivalent topological physical model, a backtracking process is initiated to extract the abnormal channel identifiers corresponding to the model. The core of this backtracking process is to analyze the contribution weight of the original complex impedance data of each sensing reference node (i.e., the intersection channel of the transmitting and receiving electrodes) to the final matching result and model reconstruction when calculating local geometric mapping features in the geometric feature extraction module and performing nearest-neighbor matching in the material classification module. First, the Euclidean distance evolution sequence calculated for each sensing reference node in the geometric feature extraction module is retrieved from the cache. And the final local geometric mapping eigenvector.

[0084] At the same time, retrieve all calculations performed in the material classification module. Distance values ​​and matching process records are recorded. For each sensing reference node i, its distortion contribution value is calculated. This is a quantitative indicator used to measure the contribution of a specific sensing channel to the final detected abnormal pattern. Its calculation relies on a preset contribution quantification algorithm, which may be based on a weighted summation of the point-by-point differences between the node impedance trajectory and the matching standard trajectory, or on the sum of the absolute values ​​of the differences between each element in the node's local geometric mapping feature vector and the corresponding element in the standard vector. In this embodiment, the specific calculation method is as follows: [The remaining text appears to be incomplete and requires further context for accurate translation.] Sequence under all frequency indices The values ​​are weighted and summed, with the weighting coefficients set according to the degree of influence of impedance changes at that frequency point on the matching results.

[0085] Assume the degree of influence is determined by comparing the impedance trajectory of node i with the standard trajectory matched in the database. The weighting coefficient is measured by the local differences at each frequency point. Defined as the normalized magnitude of local differences. The calculation formula is part of the system's preset contribution quantification algorithm. Subsequently, the distortion contribution values ​​of all nodes are calculated. This forms an array of contribution values. The system sets a preset distortion threshold. This threshold is a dynamic value set based on the statistical distribution of the contribution values ​​of all nodes in historical cleanroom detection data. Preset distortion threshold. The setting is to filter out channels that make significant contributions and eliminate noise interference. Its specific value can be determined based on the statistical distribution of the contribution values ​​of all nodes in a large amount of normal state detection data. For example, it can be set to A times the average contribution value of all nodes, or it can be set to the value that ranks in the top B percent after the contribution values ​​are sorted by size. The typical range of A is 1.5 to 3, and the typical range of B is 10% to 20%.

[0086] Iterate through the contribution value array and identify all The value exceeds the preset distortion threshold. The nodes are identified, and their identifiers are defined as polarization electrode channel combinations. These combinations include all transmitting and receiving electrode intersection channels that significantly contribute to the formation of the current anomalous signal. Next, the electrode channel information in active transmitting state is extracted from the polarization electrode channel combinations. The identifier of each selected node is parsed; this identifier is typically associated with its corresponding row transmitting electrode number in the internal data structure. And column receiving electrode serial number All selected nodes Extract them to form a row channel sequence number list (RL). Similarly, extract all selected nodes... Extract them to form a list of column channel serial numbers (CL).

[0087] Subsequently, the physical wiring mapping table of the capacitor matrix, stored in the underlying firmware or configuration file, is invoked. This table is a pre-measured data table based on the touchscreen's hardware design and physical layout, recording the absolute position coordinates of each electrode on the screen. Its creation is typically completed during the screen manufacturing stage through optical positioning or precision mechanical measurement. This mapping table is a two-dimensional lookup table, where row indices correspond to all possible row transmitting electrode numbers, and column indices correspond to all possible column receiving electrode numbers. Each table cell stores the precise physical coordinates of the electrode corresponding to that number on the touchscreen. For row electrodes, the table cell stores the Y-coordinate value of the electrode's centerline in the screen coordinate system. For column electrodes, the table cell stores the high-precision X-coordinate value of the electrode's centerline in the screen coordinate system. Based on the row channel sequence number list RL, the mapping table is queried sequentially to obtain the Y-coordinate corresponding to each row number, resulting in a set of Y-coordinates. .

[0088] Based on the column channel sequence number list CL, the mapping relationship table is queried sequentially to obtain the high-precision X coordinate corresponding to each column number, resulting in a set of high-precision X coordinates. Finally, the two-dimensional spatial coordinates of the actual substance type are calculated. The calculation method is as follows: The arithmetic mean of all Y-coordinate values ​​is calculated to obtain the center Y-coordinate. ;right The arithmetic mean of all high-precision X-coordinate values ​​is calculated to obtain the center high-precision X-coordinate. This method of calculating the arithmetic mean of the center coordinates is a simple algorithm for estimating the spatial center. It assumes that the area of ​​influence of the anomalous substance covers the intersection region of these anomalous electrodes, and its physical center can be approximated by the average of the coordinates of these electrodes. That is, the two-dimensional spatial coordinates of the actual material type calculated.

[0089] For example, the system has determined the substance to be deionized water and reconstructed the parallel RC equivalent topological physical model. The backtracking process is initiated, retrieving data from all nodes. The sequence and local geometric mapping feature vectors, along with the calculated distance values, are given. Assume the system has 320 nodes. Calculate the distortion contribution value for each node. For example, for node 1, this node has been determined to be a mutated node, and its... Given the sequence, calculate its... Assuming the preset contribution quantification algorithm is to... The sequence exceeds the mutation detection threshold Those The values ​​are accumulated. In the previous example, node 1 has 400 points exceeding [a certain threshold]. , these 400 points The values ​​are accumulated to obtain a larger value. The value, for example, 5000. For other normal nodes that were not identified as mutated nodes, their... The sequence contains very few or no points exceeding the threshold. The value is very small, for example, 10. The system obtains 320 values. An array of values.

[0090] Assuming that the distortion threshold is preset based on historical data statistics Set as all nodes Twice the average, assuming the average is 50, that is... Traversing the array, it was found that only a few nodes... For nodes exceeding 100, such as 5000 for node 1, 120 for node 5, and 110 for node 8, add these nodes to the polarization electrode channel combination. Next, strip the row and column numbers of these nodes. Assume node 1 corresponds to row 3 and column 7, node 5 to row 3 and column 8, and node 8 to row 4 and column 7. Then the row channel sequence number list RL is [3,3,4], and the column channel sequence number list CL is [7,8,7].

[0091] Then, the physical wiring mapping table of the capacitor matrix is ​​called. Querying this table: for row number 3, the Y coordinate is obtained. For row number 4, obtain the Y coordinate. .therefore For column number 7, obtain the X coordinate. For column number 8, obtain the X coordinate. .therefore Finally, calculate the center coordinates: ; The two-dimensional spatial coordinates of the deionized water deposits are calculated as follows: .

[0092] In one embodiment of the present invention, the exception report generation module is used to perform the following steps: The process involves: calling a material type encoding mapping table to convert the actual material type into a foreign object type identifier and a touch shielding security level instruction; converting two-dimensional spatial coordinates into binary data blocks; assembling the foreign object type identifier, binary data blocks, and touch shielding security level instruction according to a predefined structured data frame format to generate a structured anomaly report; after generating the structured anomaly report, the process further includes: pushing the structured anomaly report to the main processor; and having the main processor parse the structured anomaly report and, based on the two-dimensional spatial coordinates and touch shielding security level instruction, driving the touchscreen control system to perform adaptive anti-mistouch shielding on the screen area where the actual material type is located.

[0093] Specifically, upon receiving the actual substance type and two-dimensional spatial coordinates, the data stitching and report generation process is initiated. First, a substance type encoding mapping table pre-stored in the firmware is invoked. This mapping table maps the textual description of the substance type name to a unique integer type identifier. The substance type encoding mapping table is a key-value pair data structure, where the key is the substance type name string, and the value is the associated integer identifier and a multi-bit security level instruction. This mapping table is preset during the system firmware development phase based on the expected substance types to be detected. The foreign matter type identifier is set to facilitate digital system processing and classification; for example, deionized water is mapped to identifier code 1, and silicone grease is mapped to identifier code 2.

[0094] The system locates the corresponding entry for the substance's type in the mapping table and extracts its digitized foreign object type identifier. Simultaneously, each entry in this mapping table is associated with a preset touch shielding safety level instruction. This instruction is pre-set based on the potential risk level of the substance's interference with touch signals. The setting is based on a risk assessment and grading of the potential interference level of the touch signal based on the substance's physical properties such as conductivity and dielectric constant. For example, highly conductive metal dust may cause severe false touches and is set to the highest level instruction, such as binary code 111, triggering the most stringent shielding, such as completely ignoring the signal in that area. Insulating liquids such as water may be set to a lower level, such as binary code 001, triggering a milder shielding, such as reducing sensitivity by 50%. The touch shielding safety level instruction is typically a multi-bit binary control code that defines the specific behavior pattern of subsequent shielding logic.

[0095] Next, the X and Y components of the two-dimensional spatial coordinates are converted into fixed-length binary data blocks. The foreign object type identifier is then packaged and bound to this coordinate data block, along with the touch shielding security level instruction. The packaging process follows a predefined structured data frame format, which includes a frame header, length field, type identifier field, X coordinate field, Y coordinate field, security level instruction field, checksum field, and frame trailer. It should be noted that this predefined structured data frame format ensures that the main processor can parse it correctly. The frame header is typically a fixed sequence of bytes used to identify the beginning of the data frame. The length field indicates the total number of bytes in the data frame. The checksum field ensures the integrity of data transmission, typically using CRC checksum or a simple summation check.

[0096] After assembly, a serial communication controller is invoked to convert the assembled data frame into a bit stream and encapsulate it according to the physical layer and data link layer rules of the specified serial communication protocol standard, such as the asynchronous serial communication protocol UART. The selected serial communication protocol standard must be supported by the system hardware and compatible with the main processor. Its parameters, such as baud rate, data bit length, and number of stop bits, must be pre-configured consistently between the main processor and the touch detection subsystem, for example, a baud rate of 115200bps, 8 data bits, and 1 stop bit. The encapsulation process may include adding start bits and stop bits, and arranging the bytes according to the byte order required by the protocol.

[0097] Subsequently, a structured exception report data frame is pushed to the system's main processor via a hardware serial communication interface, which includes, but is not limited to, the UART TX pin. The main processor receives and parses the data frame through the corresponding serial communication interface, extracting the foreign object type identifier, coordinate information, and security level instructions. Based on this information, the main processor triggers a redraw of the interface state, such as overlaying a visual marker at the corresponding coordinate position in the graphical user interface and outputting a warning message.

[0098] Simultaneously, the main processor, based on security level instructions, drives the touchscreen control system to execute adaptive anti-mistouch shielding logic. This logic is implemented by modifying the configuration parameters of the touchscreen driver through software. The size of the rectangular area of ​​the coordinate range is usually set based on the type of substance and its estimated influence range. For example, for a water droplet, it can be set as a circular area with a radius of 10mm centered on the reported coordinates, which is then converted into a rectangular boundary in the software. The specific implementation of this logic typically involves modifying the sensitivity mapping table of the touch driver, defining a rectangular area of ​​the coordinate range around the contaminated area of ​​the screen specified in two-dimensional spatial coordinates, and reducing the touch sensitivity within this area to below a preset safety threshold. This safety threshold is an empirical value that can be determined experimentally, for example, by multiplying the original sensitivity value by a coefficient of 0.3; or by directly marking the touch signals received within this area as invalid and ignoring them, thereby completing the detection closed-loop output.

[0099] For example, the received data includes the substance type deionized water and two-dimensional spatial coordinates (106.67 mm, 166.67 mm). A pre-stored substance type encoding mapping table is consulted to find the entry corresponding to deionized water. Its digital foreign matter type identifier is extracted as 1, and the corresponding touch-screen shielding security level instruction is binary code 001, representing low risk and medium shielding. The X component (106.67 mm) and Y component (166.67 mm) of the coordinates are converted into 32-bit binary floating-point representations of data blocks. The data is assembled according to a predefined data frame format: the frame header is 0xAA 0xBB, the length field is calculated as the total number of bytes of all subsequent fields (assumed to be 20 bytes), the type identifier field is filled with 1, the X coordinate field is filled with the floating-point binary representation of 106.67, the Y coordinate field is filled with the floating-point binary representation of 166.67, the security level instruction field is filled with binary code 001, the check field is generated by summing all the aforementioned field bytes, and the frame tail is 0xCC.

[0100] After assembly, the data frames are converted into a bit stream via the serial communication controller. Assuming the UART is configured with a baud rate of 115200bps, 8 data bits, no parity, and 1 stop bit, the controller adds a start bit and a stop bit to each byte and sends them sequentially via the TX pin. The main processor receives the bit stream via the RX pin, identifies the frame header, and parses out the length, identifier code 1, coordinates (106.67, 166.67), and security level instruction 001. Based on this, the main processor draws a water droplet warning icon at coordinates (106.67, 166.67) on the graphical interface. Simultaneously, based on security level instruction 001, it drives the touchscreen control system to execute adaptive anti-mistouch blocking logic: a square area with a side length of 20mm centered at coordinates (106.67, 166.67) is defined in the software, and the touch detection sensitivity configuration value of all sensing nodes within this area is multiplied by a coefficient of 0.5, thereby reducing the sensitivity. Finally, the detection loop was completed, and a structured anomaly report was output, including the qualitative substance deionized water and the location coordinates (106.67, 166.67), and the corresponding shielding operation was implemented.

[0101] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0102] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A foreign object detection and positioning system for touch screens based on capacitance matrix spectrum analysis, characterized in that, include: The frequency sweep excitation signal generation module generates a continuous frequency sweep excitation signal sequence covering a predetermined bandwidth according to the environmental monitoring instructions, and synchronously outputs the corresponding reference clock reference. The electrode driving and network forming module is used to perform time-sequential polling driving of the transmitting electrodes of the capacitive touch screen using a continuous frequency sweep excitation signal sequence, so as to establish a local equivalent impedance network that reflects the physical state of the surface of the capacitive touch screen. The phase-sensitive demodulation module is used to capture the analog signal modulated by the local equivalent impedance network and perform phase-sensitive demodulation on the analog signal based on the reference clock reference to extract the in-phase component signal and the quadrature component signal. The complex impedance matrix construction module calculates complex plane coordinate data based on in-phase and quadrature component signals, and combines the complex plane coordinate data to construct a complex impedance characteristic matrix. The geometric feature extraction module is used to process the complex impedance feature matrix to filter out global environmental changes and extract the local geometric mapping features of corresponding abrupt change regions; The material classification module is used to compare local geometric mapping features with a pre-calibrated database of foreign object features to determine the type of substance in the mutation region. The coordinate calculation module is used to trace back the abnormal channel identifiers that contribute to the local geometric mapping features, and to use hardware array projection to calculate the two-dimensional spatial coordinates of the material type. The anomaly report generation module is used to stitch together the type of substance and two-dimensional spatial coordinates to generate a structured anomaly report that includes both qualitative and locational composite attributes.

2. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The frequency sweep excitation signal generation module is used to perform the following operations: Parse environmental monitoring commands to obtain preset scan parameter configuration files; Adjust the bias voltage of the voltage-controlled oscillator unit according to the preset scanning parameter configuration file to drive the voltage-controlled oscillator unit to output a continuous sweep frequency excitation signal sequence; The phase and frequency parameters of the continuous sweep excitation signal sequence are extracted and input into a high-precision phase-locked loop to generate a reference clock that is synchronized with the continuous sweep excitation signal sequence.

3. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The electrode driving and network forming module is used to perform the following steps: A scan timing generator is driven using a reference clock reference to generate array scan timing signals; The array scanning timing signal controls the analog multiplexed switch array to sequentially turn on the emitter electrodes of each row of the capacitive touch screen. A continuous frequency sweep excitation signal sequence is injected into the currently conducting transmitting electrode. The dielectric coupling effect of the screen surface attachments on the edge electric field is used to establish a local equivalent impedance network at the spatial intersection of the transmitting and receiving electrodes.

4. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The phase-sensitive demodulation module is used to perform the following operations: The analog signal and the reference clock reference are input to the analog multiplier for multiplication to generate a mixed signal; The mixed signal is input into a filter to remove high-frequency harmonic components and obtain low-frequency DC components; The low-frequency DC component is separated to extract the in-phase component signal representing the real part of the complex impedance and the quadrature component signal representing the imaginary part of the complex impedance.

5. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The complex impedance matrix construction module is used to perform the following steps: Using the values ​​of the in-phase component signal as the real axis coordinates and the values ​​of the quadrature component signal as the imaginary axis coordinates, the complex impedance point of a single induction node at a specific frequency is calculated. Multiple complex impedance points measured at all scanning frequencies by the same sensing node are connected in ascending order of frequency to generate a continuous impedance trajectory array. Traverse all sensing nodes across the entire screen and combine all continuous impedance trajectory arrays to generate a complex impedance characteristic matrix.

6. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The geometric feature extraction module is used to perform the following operations: Calculate the squared Euclidean distance between the complex impedance characteristic matrix and the preset cleanroom reference matrix at each frequency point to form an Euclidean distance evolution sequence; Perform a first-order difference operation on the Euclidean distance evolution sequence to eliminate the overall linear translation of the trajectory caused by changes in environmental temperature and humidity; Based on the processed trajectory data, the resonant peak and valley points are located, and the radius of curvature and the tilt angles of the major and minor semi-axis at the resonant peak and valley points are calculated to form local geometric mapping features.

7. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The material classification module is used to perform the following steps: The nearest neighbor pattern recognition algorithm is used to calculate the error distance between the local geometric mapping features and the standard feature vectors in the foreign object feature database; The minimum value in the error distance is compared with the preset error tolerance threshold. When the minimum value is less than the preset error tolerance threshold, the standard feature vector corresponding to the minimum value is determined as the matching result, and the circuit distribution attribute corresponding to the matching result is retrieved. The circuit distribution properties are analyzed to reconstruct the equivalent topological physical model, and the types of physical matter mapped by the equivalent topological physical model are output.

8. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The coordinate calculation module is used to perform the following steps: Calculate the distortion contribution value of each sensing node, and mark the nodes whose distortion contribution value exceeds the preset distortion threshold as polarization electrode channel combinations; Separate the row channel sequence number and column channel sequence number from the polarization electrode channel assembly; Call the capacitor matrix physical wiring mapping table to query the physical coordinates corresponding to the row channel sequence number and column channel sequence number; Calculate the center point of the physical coordinates to solve for the two-dimensional spatial coordinates.

9. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 1, characterized in that, The exception report generation module is used to perform the following operations: Call the material type coding mapping table to convert the actual material type into a foreign object type identification code and touch shielding security level instruction; Convert two-dimensional spatial coordinates into binary data blocks; The foreign object type identifier, binary data block, and touch shielding security level instruction are assembled according to a predefined structured data frame format to generate a structured anomaly report.

10. The touchscreen foreign object detection and positioning system based on capacitance matrix spectrum analysis according to claim 9, characterized in that, After the step of generating a structured anomaly report, the following steps are also included: Push structured anomaly reports to the main processor; and The main processor parses the structured anomaly report and, based on the two-dimensional spatial coordinates and the touch shielding security level instructions, drives the touch screen control system to perform adaptive anti-mistouch shielding on the screen area where the material type is located.

Images