Capacitive touch panel detection method

Abstract

The invention discloses a capacitive touch panel detection method, which relates to the technical field of flexible capacitive touch detection and signal processing, and comprises the following steps of: deducing deformation parameters through mutual capacitance difference and carrying out gridding correction on static coordinate mapping according to the deformation parameters to form dynamic coordinate mapping changing along with deformation; touch coordinate solution is based on a current form instead of an initial flat state, so that positioning deviation and trajectory distortion under bending, stretching and wrinkling conditions are reduced; besides, a touch mask is generated through a self-capacitance touch signal, and replacement and component decomposition are performed on mutual capacitance difference of a mask coverage area, so that deformation deduction mainly depends on global / local deformation characteristics and avoids touch local mutation, and the stability and usability of deformation parameters in the presence of touch are improved.

Description

Technical Field

[0001] This invention relates to the field of flexible capacitive touch detection and signal processing technology, and in particular to a method for detecting capacitive touch panels. Background Technology

[0002] With the development of flexible electronics and wearable devices, capacitive touch structures are used in bendable or stretchable interactive interfaces, such as the side touch areas of wearable devices, the touch areas of foldable / curved displays, and touch / interactive structures integrated into fabrics. These interfaces may bend, stretch, or wrinkle during use, causing the geometry of the touch carrier to be in a dynamic state of change.

[0003] The touch detection and positioning of existing capacitive touchpads are usually based on the preset geometric relationship of the electrode array and the planar coordinate model. Threshold judgment and position calculation (such as peak / centroid / interpolation) are performed on the distribution of capacitance changes caused by touch. When the touch surface is deformed, the relative position and coupling relationship between the electrodes change accordingly, which causes the mapping model established based on the initial geometric relationship to be inconsistent with the actual physical surface, which may cause touch position calculation errors or trajectory distortion.

[0004] On the other hand, deformation may also cause systematic changes or regional drift in capacitance measurements, which can be superimposed on local changes caused by touch, thereby affecting the reliability of touch decision under fixed threshold or conventional baseline tracking, resulting in false triggering or missed detection. Although there are technologies in existing touch controllers to compensate for baseline drift, they are usually geared towards slowly changing environments or noisy scenarios, and are difficult to cover fast, local deformation disturbances coupled with touch signals.

[0005] For deformable interfaces, existing solutions include: adding strain / inertia sensors to the touch electrode array to obtain the deformation state and perform compensation; or using the capacitance matrix change of the touch array itself to establish a correlation model between deformation and signal; and using data-driven methods to jointly identify touch and deformation features. The above solutions often lead to problems such as complex device structure, high calibration cost, and limited adaptability to random local deformation (such as fabric wrinkles). Summary of the Invention

[0006] In view of the aforementioned existing problems, the present invention is proposed.

[0007] This invention provides a capacitive touchpad detection method to solve the problems of coordinate distortion and baseline drift caused by flexible touchpad deformation, which easily leads to accidental touches and missed detections.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0009] This invention provides a capacitive touchpad detection method, applied to a flexible touchpad containing an electrode array, wherein the electrode array is used to form a mutual capacitance measurement channel and a self-capacitance measurement channel; the method includes:

[0010] Step S1: Obtain the mutual capacitance measurement value of the electrode array and form the current mutual capacitance matrix;

[0011] Step S2: When it is determined that there is no valid touch, update the reference mutual capacitance matrix based on the current mutual capacitance matrix;

[0012] Step S3: Based on the difference between the current mutual capacitance matrix and the reference mutual capacitance matrix, input the deformation inference model to obtain deformation parameters characterizing the deformation of the flexible touch panel;

[0013] Step S4: Modify the preset static coordinate mapping function according to the deformation parameters to obtain the dynamic coordinate mapping function;

[0014] Step S5: Obtain the self-capacitance touch signal of the electrode array and perform baseline compensation based on the deformation parameters;

[0015] Step S6: Map the electrode position corresponding to the compensated self-capacitance touch signal to the virtual plane coordinate system through the dynamic coordinate mapping function, and calculate and output the touch coordinates in the virtual plane coordinate system.

[0016] As a preferred embodiment of the capacitive touchpad detection method of the present invention, the deformation deduction model is a parameterized model that maps the mutual capacitance differential characteristics to deformation parameters, wherein the deformation parameters include at least one of bending curvature and stretching ratio.

[0017] As a preferred embodiment of the capacitive touch panel detection method of the present invention, the deformation parameters include global deformation components and local deformation components, wherein the local deformation components are used to characterize the spatial position and amplitude of the folds.

[0018] As a preferred embodiment of the capacitive touchpad detection method of the present invention, the modified static coordinate mapping function includes performing a nonlinear transformation on the coordinate grid based on the deformation parameters, and obtaining the mapping result of any electrode position by interpolation;

[0019] The modified static coordinate mapping function includes selecting grid nodes in the grid coordinate domain and performing triangulation, generating displacement corrections for the grid nodes based on deformation parameters and limiting the single displacement amplitude, and superimposing the displacement corrections onto the static mapping coordinates of the nodes to form the dynamic mapping coordinates of the nodes; for any electrode position, locating its triangular element and obtaining the dynamic mapping result of the electrode position based on centroid interpolation.

[0020] In a preferred embodiment of the capacitive touchpad detection method of the present invention, the determination that there is no valid touch includes: the peak value of the self-capacitance touch signal is lower than a preset threshold; the number of spatially connected regions obtained after thresholding the self-capacitance touch signal according to the preset threshold is zero; and there is no local abrupt change pattern consistent with the touch in the mutual capacitance difference.

[0021] The preset threshold is a threshold determined based on the statistical analysis of noise from self-capacitance touch signals in the absence of effective touch.

[0022] As a preferred embodiment of the capacitive touchpad detection method of the present invention, the reference mutual capacitance matrix is ​​updated by recursive filtering or exponential smoothing, and the single update amplitude is limited to suppress short-term disturbances.

[0023] As a preferred embodiment of the capacitive touch panel detection method of the present invention, the baseline compensation includes estimating the self-capacitance baseline drift of each electrode based on the deformation parameters and subtracting it from the corresponding self-capacitance touch signal;

[0024] The baseline compensation includes updating the electrode-by-electrode self-capacitance baseline drift under gating conditions without effective touch. The baseline drift includes deformation-related drift estimated by deformation parameters and slow-varying drift recursively derived from the slow-varying state. The deformation-related drift is estimated by deformation parameters using partition-shared mapping coefficients, and its single correction magnitude is limited. The slow-varying drift is recursively updated based on the deviation between the self-capacitance reading and the drift prediction when gating is enabled, and the update gain is determined by recursion using error variance. The deformation-related drift and the slow-varying drift are superimposed to form the electrode-by-electrode total drift, and subtracted from the corresponding self-capacitance touch signal to obtain the compensated self-capacitance touch signal. The partition-shared mapping coefficients are obtained by calibration under conditions without touch samples, and the coefficient amplitude is limited.

[0025] In a preferred embodiment of the capacitive touchpad detection method of the present invention, the virtual plane coordinate system is defined by the electrode geometry of the flexible touchpad under a preset reference shape.

[0026] As a preferred embodiment of the capacitive touchpad detection method of the present invention, the solution of touch coordinates includes normalizing and clustering the compensated self-capacitance touch signal, and calculating the centroid of each clustered region to output single-point or multi-point touch coordinates.

[0027] As a preferred embodiment of the capacitive touch panel detection method of the present invention, the flexible touch panel is integrated into a wearable device or electronic fabric, and its carrier material is bendable, stretchable or wrinkleable.

[0028] Through the above technical solution, the present invention can achieve at least the following beneficial effects:

[0029] To address the problem of deformation causing changes in electrode geometry and failure of fixed plane coordinate mapping, this invention uses mutual capacitance differential to deduce deformation parameters and then performs gridding correction on the static coordinate mapping accordingly, forming a dynamic coordinate mapping that changes with deformation. This allows touch coordinate calculation to be based on the current shape rather than the initial flat state, thereby reducing positioning deviation and trajectory distortion under bending, stretching, and wrinkling conditions.

[0030] To address the issues of aliasing between touch signals and deformation signals in the mutual capacitance domain, and the ease with which deformation deduction can be contaminated by touch, this invention generates a touch mask using self-capacitance touch signals and performs replacement and component decomposition on the mutual capacitance differential of the mask-covered area. This allows deformation deduction to primarily rely on global / local deformation features while avoiding local abrupt changes in touch, thereby improving the stability and usability of deformation parameters in the presence of touch.

[0031] To address the issues of overall or regional drift of self-capacitance baseline caused by deformation and unreliable fixed threshold touch decisions, this invention separates and superimposes fast deformation drift from slow environmental / device drift by deformation parameter-driven deformation-related drift estimation and gated slow drift recursive update. This makes the compensated self-capacitance touch signal closer to the local changes caused by touch, reducing the probability of false triggering and missed detection.

[0032] To address the problem that the reference mutual capacitance matrix is ​​easily corrupted under noise disturbance or misjudgment conditions, leading to subsequent differential distortion and decision drift, this invention employs a mechanism of no effective touch gating, update amplitude limitation, and abnormal channel freezing to ensure that reference updates only occur within a reliable no-touch window and suppress the cumulative impact of short-term mutations on the reference, thereby maintaining the consistency of the mutual capacitance differential in expressing deformation features.

[0033] To address the issues of spatial concentration, rapid change, and irregularity in local deformations such as folds, this invention characterizes the location and amplitude of folds through local deformation components, increases the mesh node density in the corresponding fold area, and locally reconstructs triangular meshes, thereby enhancing the degree of freedom of mapping correction locally while maintaining global stability, reducing the jump range of touch coordinates near folds, and improving interaction continuity.

[0034] To address the issues of system complexity and integration cost, this invention relies solely on the mutual capacitance and self-capacitance measurement links provided by the touch electrode array to complete deformation deduction, drift compensation, and coordinate calculation. Without introducing additional deformation sensors, a unified processing flow from deformation sensing to touch output can be formed, which is convenient for implementation in wearable devices, electronic fabrics, and other carriers using software algorithms. Attached Figure Description

[0035] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation on the scope of this application.

[0036] Figure 1 This is a flowchart of the capacitive touchpad detection method in the embodiment. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0038] All terms used in this application (including technical and scientific terms) have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted in a manner consistent with the context of this specification, and not in an idealized or overly rigid way.

[0039] Example 1:

[0040] like Figure 1 As shown, this embodiment proposes a capacitive touchpad detection method, applied to a flexible touchpad containing an electrode array, the electrode array being used to form mutual capacitance measurement channels and self-capacitance measurement channels; including the following steps:

[0041] Step S1: Obtain the mutual capacitance measurement value of the electrode array and form the current mutual capacitance matrix;

[0042] The electrode array includes transmitting and receiving electrodes. The mutual capacitance measurement channel corresponds to the coupling measurement between each transmitting electrode and each receiving electrode. A frame of mutual capacitance measurement values ​​is obtained according to a preset scanning order and arranged into a two-dimensional matrix according to the transmitting electrode number and the receiving electrode number. Each frame of the mutual capacitance matrix is ​​accompanied by a sampling timestamp, and the frame period is 2 milliseconds to 20 milliseconds. Before forming the matrix, the mutual capacitance measurement values ​​are subjected to DC bias removal and outlier removal. Outliers are determined when the change in the same matrix element between two adjacent frames exceeds 10 times the median change in the element over the most recent 64 frames.

[0043] In this embodiment, DC bias removal is performed on a per-element basis using the mutual capacitance matrix. This is achieved by estimating the average value of the element over the most recent few frames and subtracting it from the current frame. The statistical window length for the average value is an implementation parameter, defaulting to 128 frames and adjustable within the range of 32 to 256 frames. If the number of frames is insufficient during the startup phase, the statistical window length is set to the number of obtained frames, but not less than 8 frames. Specifically, the change amplitude between adjacent frames and the median multiple threshold are used to identify transient glitches and communication jitter. The multiple threshold is an implementation parameter, defaulting to 10 and adjustable within the range of 6 to 20. Matrix elements deemed abnormal are replaced by the median of the element over the most recent 8 frames. If the effective value is less than 4 frames, it is replaced by the average of the most recent effective elements in the same row and column. Furthermore, the sampling timestamp is generated from the same time base and recorded along with the mutual capacitance matrix. The frame period is set within the range of 2 to 20 milliseconds, defaulting to 8 milliseconds and adjustable within the range of 4 to 16 milliseconds to balance real-time performance and noise immunity.

[0044] Step S2: When it is determined that there is no valid touch, update the reference mutual capacitance matrix based on the current mutual capacitance matrix;

[0045] The determination of no valid touch is made if the following conditions are met simultaneously: the maximum amplitude of the self-capacitance touch signal does not exceed three times the standard deviation of the corresponding electrode baseline noise; the number of connected regions obtained by thresholding the self-capacitance touch signal is zero; and there are no local abrupt blocks in the mutual capacitance difference matrix whose area is less than 5% of the total number of electrodes and whose amplitude exceeds four times the standard deviation of the mutual capacitance difference across the entire frame. Local abrupt blocks are obtained by connected component analysis, where connectivity is defined as four adjacent nodes (top, bottom, left, and right).

[0046] Specifically, the baseline noise standard deviation is obtained by statistically analyzing the self-capacitance readings within segments without effective touch, consistent with the preprocessed output of the self-capacitance touch signal. The statistical window length is an implementation parameter, defaulting to 128 frames and adjustable within the range of 64 to 512 frames. When the statistical window contains frames that are gated as having touch, these frames are not included in the statistics and are filled with the most recent effective non-touch frame. Optionally, the multiplier threshold between the maximum amplitude and the baseline noise standard deviation is used to limit the conservatism of the non-touch gating, defaulting to 3 and adjustable within the range of 2 to 5. Values ​​below 2 are sensitive to noise and slight deformation, while values ​​above 5 result in a shorter non-touch window and slower baseline updates. Similarly, the full-frame standard deviation of the mutual capacitance difference is calculated on the difference matrix of each frame, and outlier removal consistent with the mutual capacitance matrix is ​​performed before calculation. The area of ​​a local mutation block is represented by the proportion of the number of matrix elements in its connected region to the total number of elements in the mutual capacitance difference matrix. The area threshold is set to 5% by default and can be adjusted within the range of 2% to 10%. The amplitude threshold is set to 4 by default and can be adjusted within the range of 3 to 6 to balance false positives and false negatives. Furthermore, in this embodiment, the local mutation mode is synonymous with the local mutation block, both referring to connected regions that satisfy the area and amplitude conditions.

[0047] Step S3: Based on the difference between the current mutual capacitance matrix and the reference mutual capacitance matrix, input the deformation deduction model to obtain the deformation parameters characterizing the deformation of the flexible touch panel.

[0048] The mutual capacitance difference matrix performs touch contamination suppression before being input into the deformation inference model: a touch mask is generated based on the self-capacitance touch signal, covering the touch connectivity region and its surrounding electrode neighborhood; neighborhood interpolation is used to replace the mutual capacitance difference elements covered by the mask, with the interpolation source being the nearest valid element in the same row and column outside the mask. The mutual capacitance difference matrix is ​​further decomposed into global and local components. The global component is obtained by two-dimensional low-pass smoothing, and the local component is the original difference matrix minus the global component; the deformation inference model uses the global component to estimate global deformation parameters and the local component to estimate local deformation parameters.

[0049] For example, the touch mask is generated from the touch connected region obtained by thresholding the self-capacitance touch signal. The thresholding threshold is determined by a multiple of the standard deviation of the self-capacitance baseline noise, with a default value of 3, which can be adjusted within the range of 2 to 5. An outer ring of electrode neighborhoods is obtained by performing a neighborhood expansion on the touch connected region, with the neighborhood definition consistent with the connectivity definition. Further, touch consistency is used to determine whether a local abrupt change in the mutual capacitance difference corresponds spatially to the self-capacitance touch connected region. The determination rule is that the local abrupt change block overlaps or is adjacent to the touch mask coverage area within the electrode row and column index domains, with the adjacency relationship using four adjacent elements (top, bottom, left, and right). Similarly, neighborhood interpolation replacement performs a search for the nearest valid value in the row and column directions of the mask coverage element. If valid values ​​exist in both the same row and column, the average of the two is taken; if a valid value exists in only one direction, the valid value in that direction is taken; if no valid values ​​exist in either direction, the average of the difference matrix across the entire frame is taken. Optionally, the two-dimensional low-pass smoothing is implemented using a two-dimensional moving average kernel. The kernel size is an implementation parameter, with a default value of 5, which can be adjusted within the range of 3 to 11. The boundary values ​​are copied to extend the smoothing process. The number of smoothing iterations is an implementation parameter, with a default value of 1, which can be adjusted within the range of 1 to 3 to adapt to different deformation scales.

[0050] Step S4: Modify the preset static coordinate mapping function according to the deformation parameters to obtain the dynamic coordinate mapping function;

[0051] The static coordinate mapping function maps the electrode row and column indices to two-dimensional geometric coordinates under the reference shape, with the two-dimensional geometric coordinates defined using the electrode spacing as the unit of measurement. The dynamic coordinate mapping function takes the static geometric coordinates as input and superimposes them with the displacement field determined by the deformation parameters to obtain the deformed two-dimensional geometric coordinates. The displacement field consists of grid control points, which are sampling points at equal intervals from the corners and boundaries of the electrode array, with the number of control points ranging from 12 to 64. The displacement of the control points is determined by the global deformation components, which control the overall bending and stretching, and by the local deformation components superimposed with local perturbations at the corresponding positions. The displacement of any electrode position is obtained by piecewise affine interpolation of the control point displacements, with the piecewise division using triangulation and a fixed topology.

[0052] In this embodiment, the two-dimensional geometric coordinates under the reference shape are defined with the electrode spacing as the scale unit. The electrode spacing is taken as the average distance between the centers of adjacent electrodes as the calibration value and fixed in the static coordinate mapping function. When the spacing of the electrode array in two directions is inconsistent, the two-dimensional geometric coordinates are scaled using the spacing in both directions respectively. Specifically, the control points are composed of equally spaced sampling points at the corner points and boundaries of the electrode array. The sampling interval is an implementation parameter, and the number of control points is set to be around 24 by default, which can be adjusted within the range of 12 to 64. When the number of boundary electrodes is insufficient to meet the requirements of equally spaced sampling, the control points are sparsely sampled according to the actual number of boundary electrodes, while keeping the corner points mandatory. Furthermore, the triangulation topology of the piecewise affine interpolation is fixed after the control point set is determined. Subsequently, only the displacement of the control points is updated without changing the triangle connection relationship. When local wrinkles occur and local refinement is performed, only control points are added in the corresponding region of the wrinkle and the local triangular mesh is reconstructed in that region. The triangulation of the remaining regions remains unchanged to avoid global mapping jumps. Optionally, the upper limit parameter of the single displacement amplitude limit is measured in two-dimensional geometric coordinates, with a default value of 0.5, which can be adjusted in the range of 0.2 to 2 to suppress sudden changes in control point displacement caused by short-term disturbances.

[0053] Step S5: Obtain the self-capacitance touch signal of the electrode array and perform baseline compensation based on the deformation parameters;

[0054] The self-capacitance measurement channel corresponds to the measurement of the capacitance to ground of each electrode in the electrode array. A frame of self-capacitance touch signal is acquired in a manner aligned with the timestamp of the mutual capacitance frame. When mutual capacitance and self-capacitance are sampled using time-division multiplexing, the sampling start time of the self-capacitance frame falls within 0.5 milliseconds to 5 milliseconds after the sampling of the corresponding mutual capacitance frame ends. The self-capacitance touch signal undergoes bandwidth limiting and power frequency interference removal processing before baseline compensation. The noise standard deviation after processing is obtained by statistically analyzing the most recent 128 frames of non-touch segments.

[0055] Furthermore, bandwidth limiting is implemented using digital filtering. The filter cutoff frequency is an implementation parameter, defaulted to 0.05 to 0.2 times the reciprocal of the frame period, and can be adjusted within the range of 0.02 to 0.3 times to strike a tradeoff between suppressing slow drift and preserving touch transients. Power frequency interference removal is implemented using notch filtering, with the notch center frequency set to 50 or 60 Hz, preset according to the equipment deployment area and fixed during initialization. Similarly, the no-touch segments of the noise standard deviation are determined by the no-effective-touch gating value, which is a binary signal of the implementation parameter. By default, the gating requires at least 8 consecutive no-touch frames to be included in the statistics, and can be adjusted within the range of 4 to 20 frames to reduce gating jitter. Optionally, when the timestamp alignment error between mutual capacitance and self-capacitance exceeds the threshold of the implementation parameter, the self-capacitance frame is replaced by the self-capacitance result of the previous frame with the same timestamp as the most recent mutual capacitance frame. The alignment error threshold is set to 2 milliseconds by default and can be adjusted within the range of 1 to 5 milliseconds. The replaced frame is marked as a missing frame and does not participate in the baseline noise standard deviation statistics.

[0056] Step S6: Map the electrode position corresponding to the compensated self-capacitance touch signal to the virtual plane coordinate system through a dynamic coordinate mapping function, and solve and output the touch coordinates in the virtual plane coordinate system.

[0057] The virtual planar coordinate system is consistent with the two-dimensional geometric coordinate system under the reference shape. The origin is taken as the geometric coordinate of the upper left corner electrode of the touchpad, the horizontal axis is along the increasing direction of the electrode column, and the vertical axis is along the increasing direction of the electrode row. During touch coordinate calculation, the mapped electrode points are sampled as scattered points to construct a regular grid intensity map. The intensity map is generated by bilinear rasterization of the scattered signals and then normalized in amplitude. Thresholding is performed on the normalized intensity map to obtain candidate touch areas. The area of ​​the candidate touch areas is limited to 0.2% to 8% of the total number of grids. The centroid of each candidate touch area is calculated as the touch coordinate output. When the number of candidate touch areas is greater than 1, multi-point touch coordinates are output and sorted in descending order of centroid amplitude.

[0058] Specifically, the resolution of the regular grid is an implementation parameter, defaulting to 4 grid units per electrode spacing, which can be adjusted within the range of 2 to 8 to strike a balance between computational complexity and spatial accuracy. During rasterization, boundary clipping is performed on the coordinates of scattered points mapped outside the grid domain, and the number of clippings is recorded. When the number of clippings exceeds 10% of the total number of scattered points, the frame is marked as a boundary frame. Further, amplitude normalization performs linear scaling on the intensity map of each frame, normalizing its maximum value to 1 and applying a lower limit clamp to excessively small maximum values. The lower limit is an implementation parameter, defaulting to the normalized amplitude corresponding to 3 times the standard deviation of the capacitance baseline noise, which can be adjusted within the range of 2 to 6 times to avoid amplification of pure noise frames. Optionally, the thresholding threshold is a fixed percentage within the range of 0.2 to 0.5 of the normalized maximum value, with a default of 0.3, and is chosen to be more stringent than the baseline noise standard deviation threshold to avoid weak touch and noise connectivity expansion.

[0059] For example, the area of ​​a candidate touch region is calculated based on the number of grid cells contained in the connected component after thresholding, and compared with the proportion of the connected component area to the total number of grid cells in a range of 0.2% to 8%. When the connected component area is less than 0.2%, it is judged as an isolated noise point and discarded; when the connected component area is greater than 8%, it is judged as a large area coverage and no touch coordinates are output. Furthermore, when the number of candidate touch regions is 0, the output of this frame is empty coordinates and the deformation parameters of the previous frame are retained for subsequent mapping updates; when the number of candidate touch regions exceeds the upper limit of the implementation parameters, only a few regions with the largest centroid amplitude are retained for output. The upper limit is 5 by default and can be adjusted within the range of 2 to 10 to adapt to multi-touch and noise reduction requirements.

[0060] In this embodiment, the deformation deduction model is a parameterized model that maps the mutual capacitance differential characteristics to deformation parameters, and the deformation parameters include at least one of bending curvature and stretching ratio.

[0061] The deformation inference model includes the mapping relationship from features to parameters obtained from offline calibration. The calibration process applies multiple sets of bending and stretching conditions to the flexible touch panel and collects the mutual capacitance matrix as samples under each set of conditions. For each sample, four types of features are extracted: row direction gradient distribution, column direction gradient distribution, full frame mean drift, and partition mean drift. The corresponding condition labels are recorded as the true values ​​of the deformation parameters. The mapping relationship is realized by a piecewise regression table. Each segment is jointly indexed by the bending radius interval and the stretching ratio interval. The number of indexed intervals is 4 to 20 segments.

[0062] In this embodiment, the row gradient distribution is obtained by differencing adjacent elements in the cross-capacitance difference matrix in the row direction, and the distribution vector is obtained by taking the absolute value of the difference result and statistically analyzing it according to the preset bins; the column gradient distribution is obtained in the same way in the column direction. The number of bins is an implementation parameter, with a default value of 16, which can be adjusted within the range of 8 to 64; the bin boundaries are determined by the gradient amplitude quantile points of the non-touch samples and fixed after calibration. Further, the full-frame mean drift is the mean of all elements in the cross-capacitance difference matrix, and the partition mean drift is the set of mean values ​​of the cross-capacitance difference matrix within the preset partition. The partitioning method is consistent with the partitioning function of the partition sharing coefficient described later; the number of partitions is an implementation parameter, with a default value of 16, which can be adjusted within the range of 4 to 64 to balance expressive power and calibration sample size. Similarly, the working condition label, as the true value of the deformation parameter, is directly recorded from the set values ​​when applying bending and stretching working conditions, and corresponds one-to-one with the collected cross-capacitance matrix through timestamps. The time resolution of the set value recording is by default no less than the cross-capacitance frame period.

[0063] Optionally, the segment boundaries of the bending radius interval and the stretching ratio interval are determined by the quantiles of the calibration samples in the corresponding dimensions, ensuring that the number of samples in each segment is not less than the lower limit of the implemented parameters. The lower limit of the sample number is set to 50 by default and can be adjusted within the range of 20 to 200 to balance the number of segments and the stability of the fit. Furthermore, the query of the segmented regression table is completed by first locating the interval and then reading the parameters. When the sample falls at the interval boundary, a linear transition between adjacent intervals is used to reduce the jump in deformation parameters caused by interval switching. The linear transition bandwidth is the implemented parameter, which is set to 10% of the interval width by default and can be adjusted within the range of 0% to 30%.

[0064] In this embodiment, the deformation parameters include global deformation components and local deformation components. The local deformation components are used to characterize the spatial location and amplitude of the folds.

[0065] Local deformation components are represented by several local peak clusters. Each local peak cluster is determined by the extreme point of the mutual capacitance local component and its connected region. The position of the local peak cluster is given by the electrode row and column index, and the amplitude is given by the average absolute value of the mutual capacitance local component in the connected region. When the number of local peak clusters exceeds 8, the 8 peak clusters with the largest amplitude are retained as local deformation components, and the remaining peak clusters are recorded as noise and discarded.

[0066] Specifically, extrema are obtained by comparing local neighborhoods on the mutual capacitance local components. The neighborhood range is an implementation parameter, defaulting to a 3x3 neighborhood centered on the point, which can be adjusted from 3x3 to 7x7. Extrema must satisfy the condition that their amplitude is not less than twice the full-frame standard deviation of the mutual capacitance local components, with a default multiplier of 2, adjustable from 1.5 to 4 to suppress noise peaks. Further, connected regions are expanded using extrema as seeds on the thresholding results of the mutual capacitance local components. The threshold is a fixed proportion within the range of 0.3 to 0.7 of the extrema amplitude, defaulting to 0.5, adjustable from 0.2 to 0.8 to cover different wrinkle widths. Similarly, when connected regions of multiple extrema overlap, they are merged into the same local peak cluster, with the extrema with the larger amplitude used as the representative point to ensure consistency between the statistical count of local peak clusters and subsequent selections.

[0067] In this embodiment, the modified static coordinate mapping function includes performing a nonlinear transformation on the coordinate grid based on the deformation parameters, and obtaining the mapping result of any electrode position through interpolation;

[0068] The modified static coordinate mapping function includes selecting grid nodes in the raster coordinate domain and performing triangulation, generating displacement corrections for the grid nodes based on deformation parameters and limiting the amplitude of single displacements, and superimposing the displacement corrections onto the static mapping coordinates of the nodes to form the dynamic mapping coordinates of the nodes; for any electrode position, locating its triangular element and obtaining the dynamic mapping result of the electrode position based on centroid interpolation; when the deformation parameters include local folding components, increasing the grid node density in the folded region and reconstructing the local triangular mesh so that the mapping changes caused by folds are concentrated in the local region; and constructing a continuous displacement field based on the grid node displacement using radial basis functions, smoothing the displacement corrections of the grid nodes using the continuous displacement field and superimposing them onto the static mapping coordinates of the nodes to form the dynamic mapping coordinates of the nodes.

[0069] Furthermore, the grid coordinate domain uses the electrode row and column indices as its coordinate basis, with its origin at the row and column index of the top-left corner electrode of the electrode array. The horizontal coordinate increases with the electrode column, and the vertical coordinate increases with the electrode row. The coordinate grid consists of a set of grid nodes within the grid coordinate domain. The grid nodes include corner points and boundary sampling points and can be refined in corresponding regions according to local wrinkle components. Triangulation is performed using a fixed-topology triangular mesh generation method. After generation, each electrode position can be located in a unique triangular unit within the grid coordinate domain. When the electrode position falls on the boundary of a triangular unit, a priority unit is selected according to the implementation parameters. The priority rule defaults to selecting the unit with the larger area, and it can switch between area priority and index priority. Optionally, when the triangular unit positioning fails or a degenerate triangle appears, the dynamic mapping result of the electrode position degenerates into its static mapping coordinates and is marked as a degenerate point. The degenerate point does not participate in the touch coordinate calculation of this frame but participates in the normal calculation of subsequent frames.

[0070] In one implementation, the static coordinate mapping function is written as a mapping function for the two-dimensional positions of the electrode array:

[0071] When the grid coordinates of the electrode within the array are: At that time, the corresponding static virtual plane coordinates are:

[0072] (1)

[0073] In equation (1), Indicates the first Two-dimensional grid coordinates of each electrode within the electrode array; express The horizontal coordinate components; express The vertical coordinate components; Indicates the first under the preset reference form Each electrode is mapped to two-dimensional coordinates in a virtual plane coordinate system; This indicates a preset static coordinate mapping function.

[0074] After obtaining the deformation parameters, a nonlinear transformation of the coordinate grid can be achieved by piecewise affine transformation, and the mapping result of arbitrary electrode positions can be output by interpolation, as follows:

[0075] When selecting A set of grid nodes constitutes a node set. ,node The raster coordinates are ,node The static coordinates in the virtual plane are The deformation parameter vector output by the deformation derivation model is denoted as:

[0076] ,

[0077] And convert it into nodal displacements to drive the movement of mesh nodes:

[0078] (2)

[0079] In equation (2), This represents the deformation parameter vector characterizing the deformation of the flexible touchpad; express The Each component is used to bear components such as bending curvature, stretching ratio, or local amplitude of wrinkles; The dimension of the deformation parameter vector; Represents a node Displacement correction amount in the virtual plane; This represents the nodes obtained from calibration or simulation. Deformation response matrix, used to... Mapped to .

[0080] when When affected by short-term disturbances, their amplitude is constrained to suppress sudden changes:

[0081] (3)

[0082] In equation (3), Represents the Euclidean norm; This indicates the upper limit of node displacement amplitude, used to limit the maximum displacement amount in a single mesh correction.

[0083] After the node moves, the node The dynamic virtual plane coordinates are written as:

[0084] (4)

[0085] In equation (4), Represents a node Dynamic coordinates within a virtual planar coordinate system.

[0086] When the node set Triangulation is performed within the raster coordinate domain to obtain a set of triangles. For any electrode ,exist The positioning includes triangle The centroid coordinates are then used to interpolate the points inside the triangle. The centroid weight can be obtained by the following formula:

[0087] (5)

[0088] In equation (5), Indicates electrode The weight of the centroid relative to the three vertices of the triangle; express The Each weighted component; This represents a coefficient matrix composed of the grid coordinates of the vertices of a triangle; Let represent a homogeneous vector composed of electrode grid coordinates, where:

[0089] (5.1)

[0090] In equation (5.1), Represents a node Horizontal grid coordinates; Represents a node Vertical grid coordinates; Indicates the electrode The raster coordinates are expanded into homogeneous vectors; express The inverse matrix.

[0091] When obtained Then, the dynamic virtual plane coordinates of the triangle vertices are interpolated according to weights to obtain the mapping results for arbitrary electrode positions:

[0092] (6)

[0093] In equation (6), Indicates electrode Dynamic coordinates mapped to a virtual planar coordinate system; Represents the triangle's first Dynamic virtual plane coordinates of each vertex node; express The The index of each vertex node. about As can be seen from the piecewise representation, the mapping switches affine segments at the boundary of the triangle, presenting a nonlinear transformation effect overall. Moreover, the interpolation calculation only involves matrix inversion and weighted summation, which is suitable for real-time solution.

[0094] When the deformation parameters include local fold components and the spatial location of the folds is concentrated in a certain sub-region, the node density can be increased in that sub-region and only the local triangular mesh can be reconstructed. This will concentrate the response of the node displacement to the fold amplitude within the local mesh, thereby reducing the range of coordinate jumps near the folds. This approach keeps the architecture of deformation parameter-driven mesh transformation-interpolation output mapping unchanged, and only changes the local refinement strategy of node distribution and triangulation.

[0095] For enhanced global smoothness, the interpolated displacement field can be replaced with radial basis function interpolation: treating nodal displacements as control points, the displacement field is constructed as follows:

[0096] ,

[0097] And let the dynamic mapping be:

[0098] (7)

[0099] In equation (7), Represents raster coordinates as The virtual plane displacement correction at the location; The first radial basis interpolation represents the... Vector coefficients; Represents the radial basis kernel function; Represents a node The raster coordinates. Coefficients. By interpolation constraints Solve to achieve consistency with piecewise affine mapping at nodes and smooth transition between nodes using kernel functions.

[0100] For example, the kernel function of the radial basis function monotonically decays with increasing grid coordinate distance and strictly satisfies the control point displacement constraint at the control points. To avoid coefficient instability caused by sparse control point distribution or local densification, a numerical stabilization term is introduced in the coefficient solution. The stabilization strength is an implementation parameter, with a default value of 0.000001, which can be adjusted within the range of 0.0000001 to 0.0001. Furthermore, when the number of control points introduced by local densification causes the solution scale to exceed the upper limit of the implementation parameter, corner and boundary control points are retained first, and control points are retained in the corresponding fold regions in descending order of amplitude. The upper limit is 64 by default and can be adjusted within the range of 32 to 128 to control the computational load.

[0101] It can be seen that: after each deformation parameter is obtained, the reference coordinates of the electrode and the mesh node in the virtual plane are first obtained by the static coordinate mapping relationship, which is given by equation (1); then the displacement correction of the mesh node in the virtual plane is calculated based on the deformation parameter, the node displacement is calculated according to equation (2), and the single displacement amplitude is limited according to equation (3) to suppress short-term disturbances; the node displacement is superimposed on the reference coordinates of the node to obtain the dynamic node coordinates, and the update is completed according to equation (4); then the nodes in the grid domain are triangulated and the triangular unit where any electrode is located is located. The centroid weight is obtained in the unit according to equation (5) and equation (5.1), and the dynamic node coordinates of the vertices of the triangular unit are interpolated according to equation (6) to output the dynamic mapping coordinates of the electrode in the virtual plane; when it is necessary to improve global smoothness, a continuous displacement field is constructed based on the node displacement and superimposed on the static mapping to obtain the dynamic mapping result, which is output according to equation (7) to reduce the coordinate discontinuity at the switching point of adjacent units.

[0102] Specifically, the above implementation method revolves around the generation process of the dynamic coordinate mapping function, transforming the deformation information obtained from deformation deduction into node movement of the coordinate grid, and extending the node movement to arbitrary electrode positions through interpolation. By mapping the two-dimensional position of the electrode array to a virtual plane coordinate system, the geometric changes caused by bending, stretching, or wrinkling can be reflected as the displacement correction of the nodes in the virtual plane. The piecewise affine method uses a triangular mesh to decompose the global transformation into multiple local segments. Electrode points are interpolated within their respective triangles according to the centroid weight. The computational load increases linearly with the number of nodes, facilitating real-time operation in the touch calculation link. Constraints on the displacement amplitude can suppress mapping jumps caused by short-term disturbances, and locally refined meshes can limit the influence of wrinkles to the target area, thereby reducing the drift range of touch coordinates near wrinkles. When a smoother transition is required, radial basis functions are introduced to make the displacement field continuous, which can reduce boundary discontinuities caused by the switching of adjacent segments.

[0103] In this embodiment, determining that there is no valid touch includes: the peak value of the self-capacitance touch signal is lower than a preset threshold; the number of spatially connected regions obtained after thresholding the self-capacitance touch signal according to the preset threshold is zero; and there is no local abrupt change pattern consistent with the touch in the mutual capacitance differential.

[0104] The preset threshold is a threshold determined based on the statistical analysis of noise from self-capacitance touch signals in a state of no effective touch.

[0105] In this embodiment, the updated reference mutual capacitance matrix is ​​updated using recursive filtering or exponential smoothing, and the magnitude of each update is limited to suppress short-term disturbances.

[0106] The reference mutual capacitance matrix is ​​updated using an exponential smoothing method. The update ratio of the new reference to the old reference is 0.5% to 3%, and the update is only performed when the determination of "no valid touch" is met. A single update amplitude limit is set for each matrix element, which is 0.2% of the current reference value of the element. When an element triggers the amplitude limit for 3 consecutive frames, the update of the element is frozen for 50 frames and recorded as an abnormal channel.

[0107] Similarly, during the freeze period, the abnormal channel retains its baseline mutual capacitance matrix elements without updating and still participates in the mutual capacitance differential calculation. After the freeze period ends, it is gradually unfrozen in an exponential smoothing manner. The update percentage during the unfreezing phase is an implementation parameter, defaulting to 0.5% and adjustable within the range of 0.1% to 2% to avoid abrupt changes. Furthermore, when the cumulative number of freezes for the same abnormal channel exceeds the implementation parameter threshold within 200 consecutive frames, the channel is marked as a persistent abnormality and is always treated as an invalid element in the subsequent touch contamination suppression stage. Invalid elements are replaced by the average of the valid elements in the same row and column and enter the differential and decomposition process. The cumulative freeze threshold defaults to 3 and can be adjusted within the range of 2 to 10.

[0108] In this embodiment, baseline compensation includes estimating the self-capacitance baseline drift of each electrode based on deformation parameters and subtracting it from the corresponding self-capacitance touch signal;

[0109] The baseline drift of the self-capacitance for each electrode is obtained by looking up the deformation parameters in a table. The table index consists of the global bending condition, the global tensile condition, and the number of local peak clusters. Each electrode corresponds to a set of drift entries, which are obtained by statistically analyzing the self-capacitance data collected under non-touch conditions during the offline calibration phase. After subtracting the drift, zero-point clamping is performed on the residual signal of each electrode. The clamping rule is to set the residual signal to zero when it is less than zero. Electrodes with residual signals of zero for five consecutive frames are marked as background electrodes, and background electrodes do not participate in the construction of candidate touch areas.

[0110] Baseline compensation includes updating the electrode-by-electrode self-capacitance baseline drift under gating conditions without effective touch. The baseline drift includes deformation-related drift estimated from deformation parameters and slow-varying drift recursively derived from slow-varying states. The deformation-related drift is estimated from deformation parameters using partition-shared mapping coefficients, and its single correction magnitude is limited. The slow-varying drift is recursively updated based on the deviation between the self-capacitance reading and the drift prediction when gating is enabled, and the update gain is determined by recursion using error variance. The deformation-related drift and the slow-varying drift are superimposed to form the total electrode-by-electrode drift, and subtracted from the corresponding self-capacitance touch signal to obtain the compensated self-capacitance touch signal. The partition-shared mapping coefficients are obtained through calibration under conditions without touch samples, and the coefficient amplitude is limited.

[0111] In one implementation, the electrode-by-electrode self-capacitance baseline drift can be placed between "acquiring the self-capacitance touch signal" and "normalization and clustering calculation" as a slow variable correction link linked to the deformation parameters. The self-capacitance reading is denoted as... , indicating the first The electrode at the first The self-capacitance reading at the next sampling time; Indicates the electrode index; Indicates the sampling sequence number.

[0112] To ensure that drift estimation is updated only when there is no active touch, a touchless gating variable is introduced:

[0113] (9)

[0114] In equation (9), Indicates the first The touchless gating quantity at the next sampling time; the touchless determination condition refers to the aforementioned determination logic of "the self-capacitance peak and its spatial connectivity region are below the threshold and the mutual capacitance difference does not show a touch-consistent local abrupt change mode".

[0115] In this embodiment, the no-touch gating value is consistent with the aforementioned determination logic for the absence of valid touch. A gating value of 1 indicates that updating is allowed, and a value of 0 indicates that updating is prohibited. To reduce gating jitter, a hysteresis rule is adopted for the gating value. By default, updating is allowed only if the no-touch determination is met for two consecutive frames, and updating is prohibited only if the determination is not met for two consecutive frames. This value can be adjusted within the range of 1 to 5 frames. Furthermore, when the gating value is in the prohibited update state and the duration exceeds the implementation parameter threshold, the slow drift maintains the previous valid update value and only performs deformation-related drift subtraction. The duration threshold is set to 200 frames by default and can be adjusted within the range of 50 to 1000 frames to adapt to different environmental drift rates.

[0116] 1) Linear mapping from deformation parameters to electrode-by-electrode drift:

[0117] When the deformation deduction model outputs the deformation parameter vector as At this time, the self-capacitance baseline drift introduced by deformation can be written as a linear model of partition-shared coefficients:

[0118] (10)

[0119] In equation (10), Indicates the first The electrode at the first The baseline drift estimate of self-capacitance caused by deformation at the next sampling time; Indicates the electrode The deformation-drift coefficient vector corresponding to the partition; Indicates the electrode index Partition functions mapped to partition indices; Indicates the first The deformation parameter vector at the next sampling time, the physical meaning of the deformation parameters refers to the definition in the previous steps; This indicates the transpose operation.

[0120] To reduce the number of coefficients resulting from independent modeling of each electrode, the electrode array can be divided into several partitions and shared. The zoning method can be set according to the electrode geometry, wiring grouping, or material splicing area, so that the baseline drift caused by deformation presents a similar trend within the zoning.

[0121] Specifically, the partitioning function maps electrode indices to partition indices. Partitions are obtained by equally dividing the electrode row and column indices in two directions. The number of divisions is an implementation parameter, defaulting to 4 divisions in each direction, but adjustable from 2 to 8 divisions. When the number of electrodes in the electrode array is asynchronous in the two directions, the array is independently divided according to the number of electrodes in each direction, with adjacent electrodes continuously belonging to the same partition as a constraint. Furthermore, the mapping coefficients shared by the partitions are obtained under non-touch sample conditions during the offline calibration phase, and amplitude limitations are applied after the coefficients are obtained. The upper limit of the amplitude is an implementation parameter, defaulting to the order of magnitude of 10 times the standard deviation of the capacitance baseline noise, and adjustable from 5 to 30 times to avoid coefficient amplification due to small samples.

[0122] When it is necessary to limit abnormal deformation or drift abrupt changes caused by short-term disturbances, it is possible to... Apply amplitude constraints:

[0123] (11)

[0124] In equation (11), This represents the upper limit of the deformation drift estimate, used to limit the range of baseline corrections resulting from a single deformation mapping.

[0125] 2) Incorporate slow temperature drift and other parameters into the state space and update them without touch gating:

[0126] In addition to drift caused by deformation, self-capacitance also exhibits slow-varying terms such as temperature drift, humidity drift, and device aging, which can introduce a slow-varying drift state into each electrode. And recursively in one-dimensional Kalman form under touchless gating:

[0127] (12)

[0128] In equation (12), Indicates the first The electrode at the first Slowly varying baseline drift estimator at the next sampling time; It represents the estimate at the previous sampling time; Indicates the first The electrode at the first The gain coefficient at the next sampling time; the residual term in parentheses represents the deviation of the current reading from the predicted value of slow drift after deducting deformation drift; Used to suppress erroneous updates of slow drift when touch is present.

[0129] The gain coefficient can be recursively derived from the error variance:

[0130] (12.1)

[0131] In equation (12.1), Indicates the first Variance of drift estimation error at a sampling time on each electrode; The process noise variance representing the drift state is used to characterize the rate of change of slowly varying drift. This represents the variance of the observation noise, used to characterize the noise level of the self-capacitance reading.

[0132] Optionally, the observation noise variance is obtained by squared the standard deviation of the self-capacitance baseline noise and maintained separately for each electrode. It is updated every 1000 frames by default and can be adjusted within the range of 200 to 5000 frames to track changes in noise level. The process noise variance is an implementation parameter used to characterize the rate of change of the slow drift. It is set to 0.001 times the observation noise variance by default and can be adjusted within the range of 0.0001 to 0.01 times to control the drift tracking speed. Further, the initial value of the error variance is an implementation parameter, which is set to 10 times the observation noise variance by default and can be adjusted within the range of 1 to 100 times. When the gating value is set to disable updates, the error variance accumulates and grows according to the process noise variance but does not absorb the observation residual, so as to avoid writing the touch signal into the slow drift when touch is present.

[0133] The corresponding error variance is updated as follows:

[0134] (12.2)

[0135] In equation (12.2), Indicates the first The electrode at the first The variance of the drift estimation error at each sampling time; the meanings of the other parameters are the same as in the previous formula.

[0136] 3) Generate the total drift per electrode and subtract it from the self-capacitance touch signal:

[0137] The deformation drift and the slow drift are combined into an electrode-by-electrode baseline drift:

[0138] (13)

[0139] In Equation (13), represents the estimated amount of self - capacitance baseline drift of the th electrode at the th sampling moment; and respectively represent the estimated drift amount caused by deformation and the slowly varying drift amount.

[0140] Based on this, baseline compensation is performed on the self - capacitance touch signal:

[0141] , (14)

[0142] In Equation (14), represents the self - capacitance touch signal after baseline compensation of the th electrode at the th sampling moment; and have the same meaning as above. is used for normalization, clustering, and centroid calculation, which can make the "false peaks" caused by deformation or slow - varying drift fall back near the threshold after deduction, reducing the touch - coordinate offset caused by the mis - expansion of the connected region.

[0143] When the value of is required, calibration can be performed on the samples where the touch - free gating is established, and a regularization constraint is added to the calibration target to limit the coefficient amplitude, so that abnormal amplification does not occur in different partitions when the samples are insufficient; this calibration process does not change the structure of "deformation - parameter - driven drift estimation - touch - free gating update - electrode - by - electrode deduction", but only provides a way to obtain the coefficients.

[0144] As described above, in each sampling period, the self - capacitance readings of each electrode are collected, and a gating quantity is generated based on the touch - free determination. Only when the gating permits, the drift state is updated. The gating logic corresponds to Equation (9); in the sampling allowed by the gating, first, the baseline drift amount related to deformation is estimated based on the deformation parameters, and the deformation - related drift is obtained according to Equation (10), and its amplitude is limited according to Equation (11) to suppress the mutation caused by abnormal deformation; then, the deviation between the self - capacitance reading and the drift prediction is used as the correction basis, and the slowly varying drift state is recursively updated according to Equation (12), where the update gain is obtained by recursively calculating the error variance, the gain calculation corresponds to Equation (12.1), and the error - variance update corresponds to Equation (12.2); subsequently, the deformation - related drift and the slowly varying drift are superimposed to form the total drift amount per electrode, and the total drift is obtained according to Equation (13); finally, the total drift is deducted from the original self - capacitance reading according to Equation (14) to obtain the compensated self - capacitance touch signal, and this compensation result enters the normalization, clustering, and centroid - solving link to output the touch coordinates.

[0145] Specifically, this implementation describes the chain for electrode-by-electrode self-capacitance baseline compensation, breaking down baseline changes into a deformation-related term and a slow-varying term caused by environmental and device factors. The deformation-related term converts deformation information into baseline offsets for each electrode through a partitioned, shared linear mapping, allowing overall lifting or sinking caused by bending, stretching, or wrinkling to be offset in advance. The slow-varying term is recursively derived in a slowly changing state form, absorbing deviations in the self-capacitance reading only when no touch gating is established, and ceasing updates when touch is present to avoid mistakenly writing touch signals into the baseline. The two parts are combined and subtracted from the original self-capacitance reading. The compensated signal is used for subsequent normalization and clustering, which can reduce the peak expansion range caused by drift, reduce the amplification of connectivity in the touch area, and thus reduce the offset amplitude and jump probability of touch coordinates in the virtual plane.

[0146] Example 2:

[0147] Based on Embodiment 1, in this embodiment, the virtual planar coordinate system is defined by the electrode geometry of the flexible touch panel under a preset reference shape;

[0148] In this embodiment, solving the touch coordinates includes normalizing and clustering the compensated self-capacitance touch signal, and calculating the centroid of each clustered region to output single-point or multi-point touch coordinates.

[0149] In this embodiment, the flexible touch panel is integrated into a wearable device or electronic fabric, and its carrier material is bendable, stretchable, or wrinkleable.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application 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 or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0151] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are meant to be within the scope of this application and form different embodiments. For example, all the embodiments above can be used in any combination. The information disclosed in this background section is intended only to enhance the understanding of the general background of this application and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art.

Claims

1. A capacitive touchpad detection method, applied to a flexible touchpad including an electrode array, wherein the electrode array is used to form a mutual capacitance measurement channel and a self-capacitance measurement channel; characterized in that, The method includes: Step S1: Obtain the mutual capacitance measurement value of the electrode array and form the current mutual capacitance matrix; Step S2: When it is determined that there is no valid touch, update the reference mutual capacitance matrix based on the current mutual capacitance matrix; Step S3: Based on the difference between the current mutual capacitance matrix and the reference mutual capacitance matrix, input the deformation inference model to obtain deformation parameters characterizing the deformation of the flexible touch panel; Step S4: Modify the preset static coordinate mapping function according to the deformation parameters to obtain the dynamic coordinate mapping function; Step S5: Obtain the self-capacitance touch signal of the electrode array and perform baseline compensation based on the deformation parameters; Step S6: Map the electrode position corresponding to the compensated self-capacitance touch signal to the virtual plane coordinate system through the dynamic coordinate mapping function, and calculate and output the touch coordinates in the virtual plane coordinate system.

2. The capacitive touchpad detection method according to claim 1, characterized in that, The deformation derivation model is a parameterized model that maps the mutual capacitance differential characteristics to deformation parameters, wherein the deformation parameters include at least one of bending curvature and stretching ratio.

3. The capacitive touchpad detection method according to claim 2, characterized in that, The deformation parameters include global deformation components and local deformation components, and the local deformation components are used to characterize the spatial location and amplitude of the folds.

4. The capacitive touchpad detection method according to claim 1, characterized in that, The modified static coordinate mapping function includes performing a nonlinear transformation on the coordinate grid based on the deformation parameters, and obtaining the mapping result for any electrode position through interpolation; The modified static coordinate mapping function includes selecting grid nodes in the grid coordinate domain and performing triangulation, generating displacement corrections for the grid nodes based on deformation parameters and limiting the single displacement amplitude, and superimposing the displacement corrections onto the static mapping coordinates of the nodes to form the dynamic mapping coordinates of the nodes; for any electrode position, locating its triangular element and obtaining the dynamic mapping result of the electrode position based on centroid interpolation.

5. The capacitive touchpad detection method according to claim 1, characterized in that, The determination that there is no valid touch includes: the peak value of the self-capacitance touch signal is lower than a preset threshold; the number of spatially connected regions obtained after thresholding the self-capacitance touch signal according to the preset threshold is zero; and there is no local abrupt change pattern consistent with touch in the mutual capacitance differential. The preset threshold is a threshold determined based on the statistical analysis of noise from self-capacitance touch signals in the absence of effective touch.

6. The capacitive touchpad detection method according to claim 1, characterized in that, The reference mutual capacitance matrix is ​​updated using recursive filtering or exponential smoothing, and the magnitude of each update is limited to suppress short-term disturbances.

7. The capacitive touchpad detection method according to claim 1, characterized in that, The baseline compensation includes estimating the self-capacitance baseline drift per electrode based on the deformation parameters and subtracting it from the corresponding self-capacitance touch signal; The baseline compensation includes updating the electrode-by-electrode self-capacitance baseline drift under gating conditions without effective touch. The baseline drift includes deformation-related drift estimated by deformation parameters and slow-varying drift recursively derived from the slow-varying state. The deformation-related drift is estimated by deformation parameters using partition-shared mapping coefficients, and its single correction magnitude is limited. The slow-varying drift is recursively updated based on the deviation between the self-capacitance reading and the drift prediction when gating is enabled, and the update gain is determined by recursion using error variance. The deformation-related drift and the slow-varying drift are superimposed to form the electrode-by-electrode total drift, and subtracted from the corresponding self-capacitance touch signal to obtain the compensated self-capacitance touch signal. The partition-shared mapping coefficients are obtained by calibration under conditions without touch samples, and the coefficient amplitude is limited.

8. The capacitive touchpad detection method according to claim 1, characterized in that, The virtual plane coordinate system is defined by the electrode geometry of the flexible touch panel under a preset reference shape.

9. A capacitive touchpad detection method according to claim 1, characterized in that, The calculation of touch coordinates includes normalizing and clustering the compensated self-capacitance touch signal, and calculating the centroid of each clustered region to output single-point or multi-point touch coordinates.

10. A capacitive touchpad detection method according to claim 1, characterized in that, The flexible touch panel is integrated into wearable devices or electronic fabrics, and its carrier material is flexible, stretchable, or wrinkleable.

Images