Keyboard key mis-touch recognition method and system based on pressure sensing

Abstract

The application discloses a keyboard key mis-touch identification method and system based on pressure sensing, and relates to the technical field of human-computer interaction.The method comprises the following steps: assigning different threshold coefficients to different regions of the keyboard based on short-time reference typing data, and deploying a first identification threshold.Real-time acquisition of keyboard pressure values is mapped to a two-dimensional keyboard coordinate space, hard boundary determination is completed relying on the first identification threshold, an effective pressure heat map is obtained, and keystroke pattern determination is carried out, and a first-order determination result distinguishing between mis-touch and non-mis-touch is output, and the non-mis-touch label corresponds to a main peak key position or a combination key.When the first-order determination is non-mis-touch, a single key discriminator is activated, the contact shape fingerprint vector is extracted in stages during the key pressing process, and a two-classification determination is completed, finally, the non-mis-touch reports the key code to trigger the input response, and the mis-touch does not generate a system event.The application solves the technical problem of mis-input caused by unintended contact during keyboard use, and achieves the technical effects of improving input accuracy and reducing mis-touch interference through identification.

Description

Technical Field

[0001] This invention relates to the field of human-computer interaction technology, and in particular to a method and system for identifying accidental key presses on a pressure-sensitive keyboard. Background Technology

[0002] Haptic keyboards are widely used in laptops and portable devices. Pressure sensors are the core hardware for key detection, and accurate key recognition directly affects the input experience and system stability. Key mis-touch suppression is a critical issue that urgently needs to be addressed in the field of haptic interaction. Existing pressure-sensitive keyboard detection solutions mostly use a uniform pressure threshold to determine key validity. This method has basic recognition capabilities in simple, regular typing scenarios, but it has significant shortcomings in complex haptic interaction scenarios for thin and light pressure-sensitive keyboards. The pressure applied to different areas of the keyboard varies greatly, resulting in poor adaptability of the uniform threshold. Relying solely on instantaneous pressure peaks cannot distinguish between large-area interference such as palm rest or fingertip scraping. When multiple keys are simultaneously pressed, it is difficult to differentiate between active key combinations and accidental touches from adjacent keys. The lack of dynamic feature verification throughout the entire pressing process makes it impossible to accurately identify various accidental touch behaviors, resulting in insufficient accuracy in accidental touch filtering and failing to meet the precise input control requirements of haptic keyboards. Summary of the Invention

[0003] This application provides a method and system for identifying accidental key presses on a pressure-sensitive keyboard, which solves the technical problem of accidental input caused by unintended contact during keyboard use.

[0004] The first aspect of this application provides a method for identifying accidental key presses on a keyboard based on pressure sensing. The method includes: using short-term reference typing data as a basis, deploying a first identification threshold by assigning differentiated threshold coefficients to different areas of the keyboard; collecting keyboard pressure values ​​through real-time pressure sensing and mapping them to a two-dimensional keyboard coordinate space; determining an effective pressure heatmap based on hard boundary judgment using the first identification threshold; performing pattern-based judgment on the effective pressure heatmap to obtain a first-order judgment result, wherein the first-order judgment result is either an accidental press or a non-accidental press, and a non-accidental press is marked with a main peak key or a combination key; if the first-order judgment result is a non-accidental press, activating a single-key discriminator, dividing the key press process into a pressing stage, a holding stage, and a releasing stage, extracting a contact pattern fingerprint vector and performing binary classification judgment, and outputting a key press judgment result; if the key press judgment result is a non-accidental press, reporting the key code to trigger an input response; if it is an accidental press, no system event is generated.

[0005] A second aspect of this application provides a pressure-sensing-based keyboard key mis-touch recognition system, the system comprising: a recognition threshold deployment module, which, based on acquired short-time reference typing data, deploys a first recognition threshold by assigning differentiated threshold coefficients to different areas of the keyboard; a pressure heatmap acquisition module, used to collect keyboard pressure values ​​through real-time pressure sensing and map them to a two-dimensional keyboard coordinate space, and perform hard boundary determination based on the first recognition threshold to determine an effective pressure heatmap; and a pattern-based determination result acquisition module, used to perform a process based on the effective pressure heatmap. The keystroke pattern is determined in a pattern-based manner to obtain a first-order determination result, wherein the first-order determination result is either a mis-touch or a non-mis-touch, and a non-mis-touch is marked with a main peak key or a combination key; the keystroke determination result output module, if the first-order determination result is a non-mis-touch, activates a single-key discriminator, divides the key pressing process into a pressing stage, a holding stage, and a releasing stage, extracts the contact pattern fingerprint vector, performs binary classification determination, and outputs the keystroke determination result; the keystroke determination result execution module, if the keystroke determination result is a non-mis-touch, reports the key code to trigger an input response, and if it is a mis-touch, no system event is generated.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages: This application divides the keyboard into regions based on the user's normal typing force and configures differentiated judgment criteria. It collects the pressure distribution of the entire area to construct a spatial pressure map. Combined with the overall characteristics of the pressure distribution, it completes the initial behavior differentiation. It performs weighted scoring and adjudication on multiple pressing points, and then breaks down the complete pressing action to extract dynamic contact features for secondary verification. In this way, it accurately distinguishes between unintentional touches and valid keystrokes, and identifies active combination keys. This makes the results of mis-touch recognition of pressure-sensing keyboards more accurate and reliable, and achieves the technical effect of improving input accuracy and reducing mis-touch interference through recognition. Attached Figure Description

[0007] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0008] Figure 1 This is a schematic flowchart of the keyboard key mis-touch recognition method based on pressure sensing provided in the embodiments of this application.

[0009] Figure 2 This is a schematic diagram of the pressure-sensitive keyboard key mis-touch recognition system provided in an embodiment of this application.

[0010] Figure labeling: Identification threshold deployment module 1, pressure heat map acquisition module 2, pattern judgment result acquisition module 3, keystroke judgment result output module 4, keystroke judgment result execution module 5. Detailed Implementation

[0011] This application provides a method and system for identifying accidental key presses on a pressure-sensitive keyboard, which solves the technical problem of accidental input caused by unintended contact during keyboard use.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "first," "second," etc., in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices.

[0014] Example 1, as Figure 1 As shown, a pressure-sensitive keyboard key mis-touch recognition method is described, wherein the method includes: Based on the acquired short-term benchmark typing data, a first recognition threshold is deployed by assigning differentiated threshold coefficients to different areas of the keyboard.

[0015] In this embodiment, the short-time reference is the baseline data of key pressure collected during the initial short period of typing by the user. It is used to adapt to the different pressing pressure habits of different users and serve as a personalized reference benchmark for accidental touch judgment. The first recognition threshold is a pressure judgment threshold system set for each key area of ​​the keyboard, serving as a judgment threshold for filtering out low-pressure invalid touches in the first round.

[0016] Specifically, based on the collected short-term benchmark typing data, the average peak pressure of each row and column of the keyboard is statistically analyzed, and differentiated threshold coefficients are assigned to different areas of the keyboard. Corresponding threshold coefficients are configured for the central high-sensitivity area, the edge low-sensitivity area, and special keys. The first recognition threshold is deployed using these differentiated threshold coefficients as hard boundaries. The specific implementation process of this step will be described in detail later.

[0017] The keyboard pressure value is collected in real time by pressure sensing and mapped to a two-dimensional keyboard coordinate space. Based on the first recognition threshold, hard boundary determination is performed to determine the effective pressure heat map.

[0018] In this embodiment, the two-dimensional keyboard coordinate space is a two-dimensional coordinate grid established according to the physical row and column layout of the keyboard. Each key corresponds to a unique coordinate point, used to locate the key position and map pressure data. The hard boundary is a fixed pressure threshold judgment rule. The pressure value is judged as valid if it reaches the threshold, and directly judged as invalid if it does not reach the threshold. There is no transition judgment interval. The effective pressure heatmap is the two-dimensional pressure distribution data of all key positions and corresponding pressure values ​​that meet the judgment criteria after eliminating low-pressure invalid touches.

[0019] Optionally, when a key is triggered, the pressure value of each triggered key is collected and mapped to a two-dimensional keyboard coordinate space to generate an instantaneous pressure heat map containing all triggered keys and their corresponding pressure values. Then, the instantaneous pressure heat map is combined with the first recognition threshold to complete the threshold matching and hard boundary determination of the corresponding position, and finally an effective pressure heat map is obtained. The specific implementation process of this step will also be described in detail later.

[0020] For the effective pressure heatmap, a pattern-based determination based on keystroke pattern is performed to obtain a first-order determination result, wherein the first-order determination result is either a mis-touch or a non-mis-touch, and a non-mis-touch is marked with the main peak key position or combination key.

[0021] In this embodiment, the keystroke mode is a type of press based on the number of peaks in the pressure field, the distribution of the center of gravity, and the uniformity of the pressure, used to distinguish different scenarios such as normal single key press, large-area accidental touch, and multi-key press.

[0022] In one embodiment of this application, the keystroke pattern is identified by extracting the main keystroke point and the pressure center offset from the effective pressure heat map. The position of the global maximum value is used as the main keystroke point. The single-peak concentrated mode, the wide and flat mode, and the multi-peak mode are distinguished according to the number of pressure peaks, the pressure value, and the distribution of the center of gravity. The main peak keystroke point is marked in the single-peak concentrated mode. The wide and flat mode is judged as a false keystroke and the keystroke response is suppressed. The multi-peak mode is transferred to the adjacent key competition adjudication. Finally, the first-order judgment result representing false keystrokes or non-false keystrokes marked with main peak keystroke points and combination keystrokes is output. Points with pressure values ​​exceeding the preset value are judged as strong pressure peaks. The specific implementation process of this step will also be described in detail later.

[0023] If the first-order determination result is that it is not a mis-touch, the single-key discriminator is activated. By dividing the key pressing process into the pressing stage, holding stage and releasing stage, the contact pattern fingerprint vector is extracted and binary classification is performed to output the key press determination result.

[0024] In this embodiment, the single-key discriminator is a functional module that performs secondary fine-tuning verification on the key signals that have passed the initial screening. It distinguishes between valid keystrokes and accidental keystrokes by analyzing the pressure morphology characteristics throughout the keystroke process. The contact morphology fingerprint vector is a set of feature data composed of the center of gravity trajectory, contact area change, pressure distribution pattern, and peak characteristics during the keystroke process, used to characterize the morphological features of the keystroke action.

[0025] Specifically, when the first-order judgment result contains a main peak key or a combination key, the single-key discriminator is activated. The pressure value of the complete pressure sensing process of the main peak key is used as the input. The key press process is divided into the pressing stage, the holding stage and the lifting stage, and the contact pattern fingerprint vector is extracted. Based on this vector, the binary classification judgment is completed, and the final output is the key press judgment result that distinguishes between non-misclick and misclick. The specific implementation process of this step will also be described in detail later.

[0026] If the keystroke determination result is not a mis-touch, the key code is reported to trigger an input response; if it is a mis-touch, no system event is generated.

[0027] Specifically, the corresponding processing is performed based on the keystroke determination result. If it is determined to be a non-misclick, the key code corresponding to the non-misclick key is stored in the input event queue. The key response and character input at the operating system level are triggered based on the input event queue. If it is determined to be a misclick, the key event is discarded and the trigger operation is marked as invalid. No key messages are transmitted to the host. The specific implementation process of this step will also be described in detail later.

[0028] Furthermore, the method provided in this application embodiment includes: Acquire short-term reference typing data, calculate the average peak pressure of each column and row of the keyboard, and assign differentiated threshold coefficients to different areas. The central high-sensitivity area is assigned a first high threshold coefficient, the edge low-sensitivity area is assigned a second low threshold coefficient, and special key positions are configured with independent threshold coefficients. The differentiated threshold coefficients are used as hard boundaries to deploy a first recognition threshold.

[0029] Specifically, after the keyboard enters calibration mode, user typing data is collected over a period of 10 to 30 seconds. A pressure sensor array distributed throughout the keyboard continuously samples the pressure of each key press, recording the peak pressure for each complete keystroke and the corresponding key's coordinates in the two-dimensional keyboard coordinate space. This data is then compiled into short-term benchmark typing data. During the data collection process, preferably, touch data with a peak pressure below 50 grams is considered invalid and discarded. This discard threshold can be adaptively adjusted based on the sensitivity characteristics of the pressure sensor array corresponding to the keyboard model and the user's typing force distribution. The selection principle is to ensure that the discarded data originates from slight contact due to incomplete keystrokes or environmental noise interference, while retaining the peak pressure records corresponding to complete keystrokes from press to release, thus ensuring that the benchmark data all originate from genuine and valid user typing operations.

[0030] After completing the short-term baseline typing data acquisition, the data is grouped according to the row and column dimensions of the keyboard's physical layout, based on the position coordinates of each key. An arithmetic mean is calculated for the effective peak pressure of all keys within each row to obtain the row-average peak pressure; similarly, an arithmetic mean is calculated for the effective peak pressure of all keys within each column to obtain the column-average peak pressure. For the baseline peak pressure of a single key, the arithmetic mean of the row-average peak pressure and the column-average peak pressure of the key is taken to represent the user's typical pressure level when striking that key.

[0031] Based on the standard keyboard layout, the keyboard is divided into zones. The central area, containing the 26 letter keys and adjacent punctuation keys, is designated as the central high-sensitivity zone. The outermost function key row, arrow key area, and control key area are designated as the edge low-sensitivity zone. Large or specially functioned keys such as the space bar, enter key, shift key, and backspace key are designated as special keys. For all keys in the central high-sensitivity zone, a first high threshold coefficient is multiplied by the corresponding key's baseline peak pressure to obtain the pressure judgment threshold. The first high threshold coefficient can be set to 0.7 to 0.9. For all keys in the edge low-sensitivity zone, a second low threshold coefficient is multiplied by the corresponding key's baseline peak pressure to obtain the corresponding key's pressure judgment threshold. The second low threshold coefficient can be set to 0.4 to 0.6, and the value of the first high threshold coefficient is always greater than the second low threshold coefficient.

[0032] For each specific key, the average peak pressure applied to that key in short-term benchmark typing data is calculated individually. A corresponding independent threshold coefficient is configured for each key. This independent threshold coefficient is multiplied by the key's own average peak pressure to obtain the pressure judgment threshold specific to that key. The independent threshold coefficient can be independently set within the range of 0.3 to 0.8, depending on the key size and usage scenario. The pressure judgment thresholds for all keys are mapped one-to-one according to two-dimensional coordinates, forming a first recognition threshold covering the entire keyboard. In subsequent real-time key detection, each key uses its own corresponding judgment threshold as a hard boundary for initial screening of valid press signals. For example, the benchmark parameters and judgment threshold configurations for each typical area and representative key are shown in Table 1.

[0033] Table 1: Comparison of First Recognition Threshold Configuration Parameters for Each Area of ​​the Keyboard ; By using row and column statistics and partition threshold calibration based on real user typing data, a hierarchical judgment boundary adapted to typing habits in different key areas was constructed, improving the adaptability and recognition accuracy of the initial accidental touch filtering.

[0034] Furthermore, the method provided in this application embodiment includes: Upon key press, the pressure value of the triggered key is determined through pressure sensing; the pressure value is mapped to a two-dimensional keyboard coordinate space to form an instantaneous pressure heat map, wherein the instantaneous pressure heat map includes all triggered keys and their corresponding pressure values; based on the first identification threshold, the instantaneous pressure heat map is subjected to threshold coefficient position matching and hard boundary determination to determine an effective pressure heat map.

[0035] Optionally, the keyboard's built-in pressure sensor array performs a row-by-row, column-by-column cyclic scan of the entire keyboard area at a fixed sampling frequency of 100Hz. When the sensor unit corresponding to any key detects a pressure signal higher than the detection lower limit, the key is determined to be a triggered key. Preferably, the detection lower limit corresponds to a pressure sampling value of 10 grams. This detection lower limit can be adaptively adjusted according to the signal-to-noise ratio characteristics of the pressure sensor array and the vibration and noise floor of the keyboard's operating environment. The setting principle is to ensure effective elimination of false trigger signals caused by the sensor unit's background noise and environmental vibration, while avoiding missing genuine user touch key operations. The real-time pressure value output by the sensor unit is read simultaneously. After a single scan is completed, the row and column position information and corresponding pressure values ​​of all triggered keys are summarized to obtain the current full-key pressure acquisition result.

[0036] Each key on the keyboard is pre-assigned a unique two-dimensional coordinate identifier. The horizontal and vertical axes of the two-dimensional coordinates correspond to the column and row numbers of the keyboard's physical layout, respectively, with a one-to-one correspondence. The pressure values ​​of each triggered key are matched to the corresponding two-dimensional coordinate points according to the key's row and column position. The pressure values ​​are then assigned to the corresponding coordinates and stored in the form of a two-dimensional array to form an instantaneous pressure heatmap. The instantaneous pressure heatmap stores the pressure values ​​at corresponding positions using two-dimensional coordinates as indices, completely including all currently triggered keys and their corresponding real-time pressure values. The pressure value at the coordinate of a key for which no pressure is detected is recorded as 0.

[0037] Finally, the system iterates through all coordinate points in the instantaneous pressure heatmap that have non-zero pressure values. Based on a first recognition threshold, it performs position matching and hard boundary determination of the instantaneous pressure heatmap using threshold coefficients. During position matching, based on the key region corresponding to the coordinate point, it matches the corresponding pressure determination threshold calculated using the differentiated threshold coefficient of that region, performing a point-by-point numerical comparison. Coordinate points with pressure values ​​greater than or equal to the corresponding determination threshold are identified as valid pressure points, and their coordinate information and pressure values ​​are retained. Coordinate points with pressure values ​​less than the corresponding determination threshold are identified as invalid pressure points, and the pressure values ​​at the corresponding coordinates are set to 0. All retained valid pressure points together constitute the valid pressure heatmap, which serves as input data for subsequent keystroke pattern determination.

[0038] By constructing instantaneous pressure distribution through fixed-frequency full-domain pressure scanning and mapping of two-dimensional coordinate arrays, and combining it with the first recognition threshold to complete partition threshold matching and hard boundary filtering, low-pressure invalid touches can be quickly eliminated, providing reliable basic data for subsequent refined mis-touch recognition.

[0039] Furthermore, the method provided in this application embodiment includes: Based on the effective pressure heatmap, the keystroke pattern is determined by identifying the main keystroke point and the pressure center offset of the pressure field, where the main keystroke point is the position of the global maximum value. If there is one strong pressure peak and at least one weak peak, and the pressure field center is close to the strong pressure peak, it is determined to be a single-peak concentration mode. In the single-peak concentration mode, the strong pressure peak is marked as the main peak key position, and the pressure value is determined to be a strong pressure peak if it is greater than a preset value. If there are multiple keys with approximate pressure values ​​and the pressure field center is dispersed, it is determined to be a wide and flat mode. In the wide and flat mode, key touch is mis-touch, and key response suppression is performed. If there are at least two strong pressure peaks, it is determined to be a multi-peak mode, and adjacent key competition adjudication is initiated.

[0040] Specifically, firstly, all effective pressure points in the effective pressure heatmap are read. The pressure values ​​of all effective pressure points are iterated through, and the point with the highest pressure value is selected as the primary keystroke point. The corresponding two-dimensional keyboard coordinates are recorded, with the keyboard coordinates in row and column numbers. The pressure-weighted centroid algorithm is used to calculate the global pressure field centroid coordinates. The x-coordinate and y-coordinate of each effective pressure point are multiplied by the corresponding pressure value, and the results are summed. Then, the summation in both directions is divided by the total pressure values ​​of all effective pressure points to obtain the x-coordinate and y-coordinate of the global pressure centroid. The row and column distance between the primary keystroke point coordinates and the global pressure centroid coordinates is calculated as the pressure centroid offset. Simultaneously, the distribution range of effective pressure points is statistically analyzed, serving as the basis for keystroke pattern recognition.

[0041] Using the pressure value of the primary keystroke point as the calculation benchmark, 70% of this benchmark value is taken as the preset value for determining a strong pressure peak. All effective pressure points are then iterated through again; points with pressure values ​​greater than or equal to this preset value are determined as strong pressure peaks, and effective pressure points with pressure values ​​less than this preset value are determined as weak pressure peaks. The total number of strong pressure peaks in the effective pressure heatmap is counted and used as the core basis for keystroke pattern classification.

[0042] When there is only one pressure peak, the row and column distance between the global pressure centroid coordinates and the coordinates of that pressure peak is calculated. If this spatial distance is less than 0.5 standard key units, the pressure field centroid is determined to be close to the pressure peak, and the corresponding keystroke mode is single-peak concentration mode. In single-peak concentration mode, the key corresponding to that pressure peak is marked as the main peak key, and the corresponding first-order judgment result is non-misclick.

[0043] When no pressure peak meeting the judgment criteria is found in the effective pressure heatmap, the variance of the pressure values ​​at all effective pressure points is calculated. If the variance of the pressure values ​​is less than 5% of the pressure value at the main keystroke, and the effective pressure points cover three or more consecutive keys, and the distance between the centroids of multiple local pressure points calculated by region splitting is greater than one standard key unit, and the dispersion of the global pressure centroid offset is high, then it is determined that the pressure values ​​of multiple keys are approximately the same and the centroids of the pressure field are dispersed. The corresponding keystroke mode is the wide-smooth mode. In the wide-smooth mode, it is directly judged as a keystroke mis-touch, and key response suppression is performed. The corresponding first-order judgment result is a keystroke mis-touch.

[0044] When the total number of pressure peaks is greater than or equal to two, the corresponding keystroke mode is determined to be a multi-peak mode, and the process enters the adjacent key competition adjudication stage. This adjudication stage further distinguishes between single-key inputs with adjacent peaks and active combination key presses. The specific implementation of this adjacent key competition adjudication will be described in detail in subsequent embodiments. After adjacent key competition adjudication, if a single-key input is determined to be valid, the corresponding main peak key is marked; if a combination key input is determined to be valid, the corresponding combination key is marked. Both types of results are categorized into the non-misclick category in the first-order judgment results.

[0045] Furthermore, the application scenarios and judgment logic for adjacent key competition adjudication under multi-peak mode will be explained below with three typical examples.

[0046] Group 1: Bimodal scenario resulting from adjacent finger position offset. The user intends to press the "J" key in the central high-sensitivity area, but because the finger's contact position is biased towards the upper right corner of the key, the pressure sensor array simultaneously detects pressure signals above the detection lower limit for both the "J" key and the "U" key above it. Furthermore, the pressure values ​​of both keys reach the preset value for pressure peak judgment, forming two pressure peaks. During the adjacent key competition adjudication process, the spatial distances of the coordinates of the "J" and "U" keys relative to the global pressure field centroid are calculated separately, and a comprehensive score is calculated by weighting the pressure value with the spatial distance. Since the actual contact center of the finger is still within the key surface area of ​​the "J" key, and the global pressure field centroid coordinates are closer to the "J" key, the comprehensive score of the "J" key is significantly higher than that of the "U" key. That is, the difference in their comprehensive scores is greater than the preset difference. The adjudication result marks the "J" key with the highest comprehensive score as the main peak key, and the corresponding first-order judgment result is non-misclick, and it is judged as a valid single-key input.

[0047] Group 2: User-initiated key combination operation scenario. The user intends to press the "Ctrl" and "C" keys simultaneously to perform a copy operation. The pressure sensor array detects pressure signals above the detection lower limit for both the "Ctrl" and "C" keys, and the pressure values ​​of both keys reach the preset value for pressure peak judgment, forming two pressure peaks. During the adjacent key competition adjudication process, the spatial distances of the coordinates of the "Ctrl" and "C" keys relative to the centroid of the global pressure field are calculated separately, and a comprehensive score is calculated by weighting the pressure value and spatial distance. Since the user presses these two keys simultaneously with approximately equal force, the centroid of the global pressure field is located in the area between the two keys, and the comprehensive scores of the two keys are similar. The difference between their comprehensive scores is less than or equal to the preset difference value. The adjudication result is determined to be a valid key combination input, the corresponding first-order judgment result is non-misclick, and the key combination is marked as "Ctrl+C".

[0048] The third group: Multi-peak scenarios caused by pressing on the edges of wide keys. The user intends to press the spacebar, but the finger's contact point is located on the edge of the spacebar, simultaneously touching the spacebar and the adjacent "Alt" or "B" key, creating two strong pressure peaks. During the competition between adjacent keys, a comprehensive score is calculated by weighting the pressure value and spatial distance. Preferably, the preset difference can be set to 15% of the maximum comprehensive score. This preset difference can be adaptively adjusted according to the physical spacing of the keyboard keys and the spatial resolution of the pressure sensor array. The setting principle is to ensure that when the user actively performs a combination key operation, the comprehensive scores of each key are similar, thus determining it as a combination key; and when a single key press causes adjacent keys to be subjected to additional pressure, the comprehensive scores of each key differ significantly, thus determining it as a valid single key input. Because the space bar's surface size is significantly larger than that of adjacent ordinary keys, even if the finger touches the edge, the global pressure field's center of gravity is still likely to be closer to the space bar. The space bar's overall score is significantly higher than that of adjacent keys, and the difference between their overall scores is greater than the preset difference. The adjudication result marks the space bar with the highest overall score as the main peak key, and the corresponding first-order judgment result is non-misclick, and it is judged as a valid single-key input.

[0049] By statistical analysis of pressure field peak values, calculation of center of gravity offset, and analysis of distribution characteristics, the automatic classification of keystroke patterns can be achieved. This can quickly intercept large-area, smooth accidental keystrokes and divert different keystroke scenarios, thereby improving the processing efficiency and initial judgment accuracy of accidental keystroke identification.

[0050] Furthermore, the method provided in this application embodiment includes: In the adjacent key competition adjudication, the spatial distance between the pressure value of each pressure peak key and the centroid of the pressure field is determined; a comprehensive score is calculated by weighting the pressure value by the spatial distance. If the difference in the comprehensive score of each pressure peak key is less than or equal to a preset difference, it is judged as a combination key; if the difference in the comprehensive score is greater than the preset difference, the key with the highest comprehensive score is marked as the main peak key.

[0051] Specifically, following the steps of the aforementioned embodiments, the multi-peak mode determination process, after the effective pressure heat map is determined to be in a multi-peak mode and enters the adjacent bond competition adjudication stage, the adjacent bond competition adjudication is executed in the following specific manner.

[0052] After entering the adjacent key competition adjudication process, the key coordinates and pressure values ​​corresponding to all pressure peaks in the effective pressure heatmap are retrieved, and the global pressure field centroid coordinates calculated based on the effective pressure heatmap are retrieved simultaneously. For each pressure peak key, using the row and column units of the two-dimensional keyboard coordinates as the measurement benchmark, the straight-line distance between the coordinates of the pressure peak key and the centroid coordinates of the pressure field is calculated using the Euclidean distance calculation method between two points, thus obtaining the spatial distance between the corresponding pressure peak key and the centroid of the pressure field. The pressure value and spatial distance parameters corresponding to each pressure peak key are stored one by one.

[0053] Next, a weighted calculation method based on pressure value and spatial distance is used to obtain the comprehensive score for each pressure peak key. The calculation rule for the comprehensive score is to directly multiply the pressure value of a single pressure peak key by the corresponding spatial distance value; the product is the comprehensive score for that pressure peak key. The same calculation is performed on all pressure peak keys to obtain the comprehensive score value for each key. The original quantization units are retained during the calculation process, without the need for additional normalization or dimension conversion.

[0054] The preset difference can be set as a fixed absolute value or as a relative proportion. Preferably, for example, the preset difference can be set to 15% of the highest comprehensive score among all pressure peaks. This preset difference can be adaptively adjusted according to the physical spacing of the keyboard keys and the spatial resolution of the pressure sensor array. The setting principle is to ensure that when the user actively performs a combination key operation, the comprehensive scores of each pressure peak key are similar, thus determining it as a combination key. When a single key press causes adjacent keys to be pressed, the comprehensive scores of each pressure peak key are significantly different, thus determining it as a valid single key input.

[0055] For scenarios with two or more pressure peaks, the pairwise differences of the comprehensive scores for all pressure peak keys are calculated. If the difference between any two sets of comprehensive scores is less than or equal to a preset difference, it indicates that the pressure distribution of each pressure peak key is balanced, and the center of gravity of the global pressure field is located in the boundary area of ​​each pressure peak key. The corresponding identification result is a combination key, and all keys corresponding to pressure peaks are marked as valid combination keys. If the difference between any set of comprehensive scores is greater than the preset difference, it is determined to be a scenario of adjacent key triggering caused by key position offset. In this case, the key with the highest comprehensive score is marked as the main peak key, and the signals corresponding to other pressure peaks are determined as incidental interference signals and are not included in the scope of valid keys.

[0056] By using a quantitative adjudication mechanism that weights pressure value and spatial distance, combined with a judgment rule that compares all pairs of data, it can accurately distinguish between active combination buttons and accidental touches of adjacent buttons in multi-peak pressing scenarios, thereby improving the recognition accuracy of multi-button synchronous pressing scenarios.

[0057] Furthermore, the method provided in this application embodiment includes: If a main peak key or combination key exists, the single key discriminator is activated. The single key discriminator takes the pressure value of the pressure sensing process of the main peak key as input, performs pressure sensing in stages on the key pressing process, performs binary classification judgment under contact morphology feature vector extraction, and determines the key press judgment result, wherein the key press judgment result is either non-misclick or misclick.

[0058] Specifically, based on the first-order judgment result output process described above, when the first-order judgment result is non-misclick and the result includes the main peak key or combination key, the single-key discriminator is activated. The single-key discriminator is a functional module that performs refined misclick verification throughout the entire single-key press process. Its core function is to perform a secondary verification of the key signals that passed the initial screening. By analyzing the detailed pressure pattern characteristics throughout the key press, it eliminates abnormal touches not identified in the initial screening, further reducing the probability of missed misclicks. For combination key scenarios, the single-key discriminator will initiate a verification process for each key included in the combination key, completing the independent judgment of each key one by one.

[0059] The single-key discriminator uses continuous pressure sampling data of the complete pressure sensing process of the main peak key as input, and retrieves the complete pressure sensing process data of each key in the combination key as input. The specific implementation method of the phased pressure sensing process of the key press will be described in detail in the corresponding embodiments later; the specific implementation method of contact morphology feature vector extraction will also be described in detail in the corresponding embodiments later.

[0060] After completing the stage division and feature extraction, the final key discriminator performs a category-by-category binary classification and comprehensive decision based on the contact morphology feature vector, and finally outputs the key strike judgment result, which is divided into two categories: non-mistaken touch and mistaken touch.

[0061] By adding a secondary discrimination module after the initial screening in the hierarchical recognition architecture, fine-grained verification is carried out across all scenarios of single keys and combination keys, thereby improving the accuracy of the final judgment result while ensuring the overall recognition efficiency.

[0062] Furthermore, the method provided in this application embodiment includes: The key press process is divided into a pressing phase, a holding phase, and a lifting phase. The pressing phase is defined as the point from contact with the key surface to the point of maximum pressure, the holding phase is defined as the stable range near the point of maximum pressure, and the lifting phase is defined as the point from the point of maximum pressure to the fingertip leaving the key surface.

[0063] In one embodiment, firstly, complete pressure timing data of the main peak key position from the moment the fingertip contacts the key surface to the moment the fingertip completely leaves the key is collected. The pressure value corresponding to each moment is recorded at a fixed sampling frequency, forming a continuous pressure change curve. For example, the moment when the pressure value first rises to the sampling value corresponding to 10 grams of force is marked as the contact start moment, and the moment when the pressure value falls back below the sampling value corresponding to 10 grams of force is marked as the contact end moment. These contact start and contact end thresholds can be adaptively adjusted according to the signal-to-noise ratio characteristics of the pressure sensing array and the vibration noise floor of the keyboard's operating environment. The setting principle is to accurately capture the start and end moments of substantial contact between the fingertip and the key surface while effectively eliminating noise interference. The global maximum value of the pressure value in the pressure timing curve is located, and the moment corresponding to this value is marked as the maximum pressure point.

[0064] Starting from the moment of initial contact and ending at the moment corresponding to the maximum pressure point, this time period is defined as the pressing phase, corresponding to the complete process from the fingertip touching the key surface to the peak pressure. Preferably, the continuous time period centered on the maximum pressure point, where the pressure value remains above 90% of the maximum pressure value, is defined as the holding phase. The threshold for this holding phase can be adaptively adjusted according to the mechanical characteristics of the key and the keystroke speed distribution of the user group. The setting principle is to identify a sufficient peak stability interval for subsequent feature extraction during normal keystrokes, and to automatically switch to the fallback judgment rule of a single sampling moment in extremely fast click scenarios. If there is no continuous stable interval that meets the duration requirement, the single sampling moment at the maximum pressure point is taken as the holding phase, corresponding to the instantaneous state when the fingertip pressure reaches its peak. Starting from the moment corresponding to the maximum pressure point and ending at the moment of contact termination, this time period is defined as the lifting phase, corresponding to the complete process from the fingertip relaxing from the peak force to leaving the key surface.

[0065] By dividing the button press into three stages using pressure-time sequence feature nodes, and supplementing the fallback rules for extreme point-press scenarios, a unified temporal segmentation benchmark is provided for subsequent multi-dimensional feature extraction, ensuring the consistency and comparability of feature extraction.

[0066] Furthermore, the method provided in this application embodiment includes: For each stage, contact morphology features are extracted to form a contact morphology fingerprint vector, and each feature category is assigned and judged. The feature categories include center of gravity trajectory, contact area change, pressure distribution pattern and peak features. The center of gravity trajectory is the two-dimensional displacement trajectory of the pressure center of gravity in the bond surface. The contact area change is the expansion or contraction curve of the contact area over time. The pressure distribution pattern is the skewness and anisotropy of the pressure distribution. The peak features are the maximum pressure value and the slope of pressure rise and fall.

[0067] Optionally, feature extraction is performed separately for the press, hold, and release phases. The feature categories include four types: center of gravity trajectory, contact area change, pressure distribution pattern, and peak features. All extracted features are concatenated in a fixed order to form a contact pattern fingerprint vector. The feature threshold range for normal keystrokes is obtained through statistical calibration of pre-collected standard keystroke samples. The judgment process performs binary classification judgment for each feature category, and then performs a comprehensive decision based on the judgment results of each category.

[0068] During the center-of-gravity trajectory extraction process, the pressure-weighted center-of-gravity coordinates within the key surface area are calculated at each sampling time. The center-of-gravity coordinates of the entire process are connected in chronological order to form a two-dimensional displacement trajectory. The total displacement length and maximum offset distance of the trajectory are statistically analyzed as corresponding feature terms. This two-dimensional displacement trajectory is marked with "start" node, "maximum pressure" node, and "end" node on its extension path within the key surface. These three nodes correspond to the contact start time, the time corresponding to the maximum pressure point, and the contact termination time, respectively, thus fully presenting the entire spatial displacement process of the fingertip pressure center of gravity within the key surface during a single key press.

[0069] During category-by-category determination, if the feature value falls within the threshold range of normal keystrokes, the category is determined to conform to normal keystroke characteristics; otherwise, it is determined to be abnormal. In the process of extracting contact area changes, for example, the number of sensor units with pressure values ​​higher than 20 grams is counted at each sampling time. This statistical threshold can be adaptively adjusted according to the sensitivity characteristics of the pressure sensor array and the ambient noise floor. The principle is to include only sensor units with substantial contact contributions in the contact area statistics, excluding invalid counts caused by noise or weak vibrations. The changes in quantity are arranged chronologically to form an expansion and contraction curve of the contact area over time. The area increase rate, peak area, and area decrease rate are extracted as corresponding feature terms, and the binary classification determination of the corresponding category is completed simultaneously.

[0070] In the pressure distribution pattern extraction process, the distribution skewness value is calculated based on the two-dimensional pressure distribution on the key surface at the peak of the key press. Simultaneously, the variances of the pressure distribution in the horizontal and vertical directions are calculated separately. The ratio of the variances in the two directions characterizes the anisotropy of the pressure distribution. The skewness and anisotropy values ​​are used as corresponding feature terms to complete the binary classification of this category. In the peak feature extraction process, the maximum pressure value throughout the entire process is extracted, and the pressure rise slope during the pressing phase and the pressure fall slope during the lifting phase are calculated. These three values ​​are used as corresponding feature terms to complete the binary classification of this category.

[0071] After completing all feature extraction and binary classification, the number of categories matching the normal keystroke characteristics is counted. Preferably, a majority voting mechanism is used for comprehensive judgment, that is, the binary classification results of each feature category are counted, and the judgment result of the majority category is used as the comprehensive decision output. Alternatively, a weighted voting mechanism can be used, assigning different weights to the distinction between normal keystrokes and accidental keystrokes based on the feature of each category, and then summarizing the judgment. Feature categories with higher distinction have a higher weight ratio in the comprehensive decision. For example, if the number of categories matching the normal keystroke characteristics exceeds 70% of the total number of categories, it is finally judged as a non-accidental keystroke; otherwise, it is judged as an accidental keystroke. This ratio threshold can be adaptively adjusted according to the keyboard model and the accidental keystroke sensitivity requirements of the application scenario. In scenarios with high accidental keystroke sensitivity requirements, the ratio threshold can be appropriately reduced to improve the accidental keystroke capture rate, and in scenarios with low accidental keystroke sensitivity requirements, the ratio threshold can be appropriately increased to reduce false judgments.

[0072] By combining and extracting multi-dimensional contact morphological features and using a category-by-category binary classification mechanism, the keystroke action features are comprehensively characterized from temporal and spatial dimensions, effectively improving the ability to identify and distinguish abnormal accidental touches.

[0073] Furthermore, the method provided in this application embodiment includes: Based on the keystroke determination result, if it is determined to be a non-misclick, the key code of the non-misclicked key is added to the input event queue; based on the input event queue, the key response and character input at the operating system level are triggered; if it is determined to be a misclick, the key event is discarded and the current trigger is marked as invalid, and no key message is sent to the host.

[0074] In one embodiment, the keyboard control unit receives the keystroke determination results output by the single-key discriminator in real time, and synchronously retrieves the pre-stored standard key codes for each key corresponding to the currently triggered event. The standard key codes adopt the USB HID standard key code encoding system and are bound and stored one-to-one with each physical key. The control unit enters the corresponding processing branch according to the category of the keystroke determination result, and performs valid input reporting or accidental touch interception operations respectively.

[0075] When the keystroke determination result is not a misclick, the valid keycode is enqueued. In single-key scenarios, the standard keycode corresponding to the main peak key is directly written to the tail of the input event queue. In combination key scenarios, the standard keycodes corresponding to all keys in the combination key are written to the input event queue in the order of the occurrence of the pressure peak of each key. The input event queue adopts a first-in-first-out storage structure, polling and dequeuing with a fixed scan period of 1ms, and outputting keycode events strictly according to the writing order to ensure that the key response order is consistent with the actual keystroke timing of the user.

[0076] Keycode events in the input event queue are dequeued sequentially according to polling order. The keyboard control unit encapsulates the keycode data into standard key data packets according to the general human-machine interface device communication protocol specification, and sends them to the host terminal via wired connection or wireless communication link. After receiving the key data packets, the host operating system completes keycode parsing, triggers the corresponding level of key response and character input operation, and completes the complete normal key input process.

[0077] When a keystroke is determined to be a mis-touch, the keyboard control unit directly discards all key data corresponding to the current trigger event, marks the trigger as an invalid event, and stores it in the local event log. This log can be used for subsequent statistical analysis of mis-touch trigger patterns and iterative optimization of identification parameters. This process does not generate any standard key data packets, nor does it send any key messages related to this trigger to the host. The host will not generate a corresponding key event response, thus achieving low-level interception of mis-touch events.

[0078] By implementing a dual-branch traffic splitting process with clearly defined coding standards and time-series scheduling, combined with the event reporting mechanism of standard communication protocols, the system can accurately intercept accidental touch events while maintaining compatibility with general host interaction logic. At the same time, invalid event data is retained to support subsequent optimization, ensuring the stability of the normal input process and the iterability of the recognition system.

[0079] In summary, the pressure-sensing-based keyboard key mis-touch recognition method provided in this application has the following technical effects: This application constructs a first recognition threshold by collecting short-time reference typing data, configuring differentiated threshold coefficients in different partitions, collecting pressure data to generate an instantaneous pressure heatmap, and obtaining an effective pressure heatmap through hard boundary judgment. Based on the pressure field characteristics, it divides the keystroke patterns to complete the first-order judgment. Non-misclick signals are subjected to phased feature extraction and binary classification verification. Finally, based on the keystroke judgment results, it is processed by diversion and layered filtering of various misclick signals, accurately distinguishing between valid key presses and misclicks, improving the accuracy of keyboard pressure-sensing misclick recognition, and achieving the technical effect of improving input accuracy and reducing misclick interference through recognition.

[0080] Example 2, as Figure 2 As shown, based on the same inventive concept as in Embodiment 1 above, this application provides a pressure-sensitive keyboard key mis-touch recognition system, the system comprising: The identification threshold deployment module 1, based on the acquired short-time reference typing data, deploys a first identification threshold by assigning differentiated threshold coefficients to different areas of the keyboard.

[0081] The pressure heat map acquisition module 2 is used to collect keyboard pressure values ​​through real-time pressure sensing and map them to a two-dimensional keyboard coordinate space, and to determine the effective pressure heat map based on the first identification threshold for hard boundary determination.

[0082] The pattern determination result acquisition module 3 is used to perform pattern determination based on keystroke pattern for the effective pressure heat map to obtain a first-order determination result, wherein the first-order determination result is a mis-touch or a non-mis-touch, and the non-mis-touch is marked with the main peak key position or combination key.

[0083] The keystroke determination result output module 4, if the first-order determination result is non-misclick, activates the single-key discriminator, divides the key pressing process into the pressing stage, holding stage and lifting stage, extracts the contact pattern fingerprint vector and performs binary classification determination, and outputs the keystroke determination result.

[0084] The keystroke determination result execution module 5 reports the key code to trigger an input response if the keystroke determination result is not a mis-touch, and does not generate a system event if the keystroke determination result is a mis-touch.

[0085] Furthermore, the identification threshold deployment module 1 is used to perform the following steps: Acquire short-term reference typing data, calculate the average peak pressure of each column and row of the keyboard, and assign differentiated threshold coefficients to different areas. The central high-sensitivity area is assigned a first high threshold coefficient, the edge low-sensitivity area is assigned a second low threshold coefficient, and special key positions are configured with independent threshold coefficients. The differentiated threshold coefficients are used as hard boundaries to deploy a first recognition threshold.

[0086] Furthermore, the pressure heat map acquisition module 2 is used to perform the following steps: Upon key press, the pressure value of the triggered key is determined through pressure sensing; the pressure value is mapped to a two-dimensional keyboard coordinate space to form an instantaneous pressure heat map, wherein the instantaneous pressure heat map includes all triggered keys and their corresponding pressure values; based on the first identification threshold, the instantaneous pressure heat map is subjected to threshold coefficient position matching and hard boundary determination to determine an effective pressure heat map.

[0087] Furthermore, the pattern determination result acquisition module 3 is used to perform the following steps: Based on the effective pressure heatmap, the keystroke pattern is determined by identifying the main keystroke point and the pressure center offset of the pressure field, where the main keystroke point is the position of the global maximum value. If there is one strong pressure peak and at least one weak peak, and the pressure field center is close to the strong pressure peak, it is determined to be a single-peak concentration mode. In the single-peak concentration mode, the strong pressure peak is marked as the main peak key position, and the pressure value is determined to be a strong pressure peak if it is greater than a preset value. If there are multiple keys with approximate pressure values ​​and the pressure field center is dispersed, it is determined to be a wide and flat mode. In the wide and flat mode, key touch is mis-touch, and key response suppression is performed. If there are at least two strong pressure peaks, it is determined to be a multi-peak mode, and adjacent key competition adjudication is initiated.

[0088] Furthermore, the pattern determination result acquisition module 3 is used to perform the following steps: In the adjacent key competition adjudication, the spatial distance between the pressure value of each pressure peak key and the centroid of the pressure field is determined; a comprehensive score is calculated by weighting the pressure value by the spatial distance. If the difference in the comprehensive score of each pressure peak key is less than or equal to a preset difference, it is judged as a combination key; if the difference in the comprehensive score is greater than the preset difference, the key with the highest comprehensive score is marked as the main peak key.

[0089] Furthermore, the keystroke determination result output module 4 is used to perform the following steps: If a main peak key or combination key exists, the single key discriminator is activated. The single key discriminator takes the pressure value of the pressure sensing process of the main peak key as input, performs pressure sensing in stages on the key pressing process, performs binary classification judgment under contact morphology feature vector extraction, and determines the key press judgment result, wherein the key press judgment result is either non-misclick or misclick.

[0090] Furthermore, the keystroke determination result output module 4 is used to perform the following steps: The key press process is divided into a pressing phase, a holding phase, and a lifting phase. The pressing phase is defined as the point from contact with the key surface to the point of maximum pressure, the holding phase is defined as the stable range near the point of maximum pressure, and the lifting phase is defined as the point from the point of maximum pressure to the fingertip leaving the key surface.

[0091] Furthermore, the keystroke determination result output module 4 is used to perform the following steps: For each stage, contact morphology features are extracted to form a contact morphology fingerprint vector, and each feature category is assigned and judged. The feature categories include center of gravity trajectory, contact area change, pressure distribution pattern and peak features. The center of gravity trajectory is the two-dimensional displacement trajectory of the pressure center of gravity in the bond surface. The contact area change is the expansion or contraction curve of the contact area over time. The pressure distribution pattern is the skewness and anisotropy of the pressure distribution. The peak features are the maximum pressure value and the slope of pressure rise and fall.

[0092] Furthermore, the keystroke determination result execution module 5 is used to perform the following steps: Based on the keystroke determination result, if it is determined to be a non-misclick, the key code of the non-misclicked key is added to the input event queue; based on the input event queue, the key response and character input at the operating system level are triggered; if it is determined to be a misclick, the key event is discarded and the current trigger is marked as invalid, and no key message is sent to the host.

[0093] The pressure-sensitive keyboard key mis-touch recognition system provided in this embodiment of the invention can execute the pressure-sensitive keyboard key mis-touch recognition method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0094] Although this application makes various references to certain modules in the system according to the embodiments of this application, any number of different modules can be used and run on user terminals and / or servers. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional unit are only for easy distinction between each other and are not used to limit the scope of protection of this invention.

[0095] The specific embodiments described above do not constitute a limitation on the scope of protection of this application. Those skilled in the art should understand that various modifications, combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the scope of protection of this application. In some cases, the actions or steps described in this application can be performed in a different order than that shown in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for identifying accidental key presses on a keyboard based on pressure sensing, characterized in that, The method includes: Based on the acquired short-term benchmark typing data, a first recognition threshold is deployed by assigning differentiated threshold coefficients to different areas of the keyboard; The keyboard pressure value is collected in real time by pressure sensing and mapped to a two-dimensional keyboard coordinate space. Based on the first recognition threshold, hard boundary determination is performed to determine the effective pressure heat map. For the effective pressure heatmap, a pattern-based determination based on keystroke pattern is performed to obtain a first-order determination result, wherein the first-order determination result is either a mis-touch or a non-mis-touch, and a non-mis-touch is marked with the main peak key position or combination key. If the first-order determination result is that it is not a mis-touch, the single-key discriminator is activated. By dividing the key pressing process into the pressing stage, holding stage and lifting stage, the contact pattern fingerprint vector is extracted and binary classification is performed to output the key press determination result. If the keystroke determination result is not a mis-touch, the key code is reported to trigger an input response; if it is a mis-touch, no system event is generated.

2. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 1, characterized in that, Deploy the first identification threshold, including: Acquire short-term baseline typing data, calculate the average peak pressure of each column and row of the keyboard, and assign differentiated threshold coefficients to different areas. The central high-sensitivity area is assigned the first high threshold coefficient, the edge low-sensitivity area is assigned the second low threshold coefficient, and special key positions are configured with independent threshold coefficients. The first identification threshold is deployed using the differentiated threshold coefficient as a hard boundary.

3. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 1, characterized in that, Perform hard boundary determination and determine the effective pressure thermogram, including: When triggered by a button, the pressure value of the triggered button is determined through pressure sensing; The pressure value is mapped to a two-dimensional keyboard coordinate space to form an instantaneous pressure heat map, wherein the instantaneous pressure heat map contains all triggered keys and their corresponding pressure values; Based on the first identification threshold, the position matching of the threshold coefficient and the determination of hard boundaries are performed on the instantaneous pressure heat map to determine the effective pressure heat map.

4. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 1, characterized in that, Perform pattern-based determination based on keystroke patterns, including: Based on the effective pressure thermogram, the keying pattern is determined by identifying the main keying point of the pressure field and the offset of the pressure center of gravity, wherein the main keying point is the position of the global maximum value. If there is a strong pressure peak and at least one weak peak, and the center of gravity of the pressure field is close to the strong pressure peak, it is determined to be a single-peak concentration mode. In the single-peak concentration mode, the strong pressure peak is marked as the main peak value key, and the pressure value is determined to be a strong pressure peak when it is greater than a preset value. If multiple keys have similar pressure values ​​and the center of gravity of the pressure field is dispersed, it is determined to be a wide and smooth mode. In the wide and smooth mode, key response suppression is implemented when keys are accidentally pressed. If there are at least two strong pressure peaks, it is determined to be a multi-peak mode, and the process enters the adjacent key competition adjudication.

5. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 4, characterized in that, In the adjacency bond competition adjudication, the spatial distance between the pressure value of each pressure peak bond and the centroid of the pressure field is determined. A comprehensive score is calculated by weighting the pressure value by the spatial distance. If the difference in the comprehensive score of each pressure peak key is less than or equal to the preset difference, it is judged as a combination key. If the difference in the overall score is greater than the preset difference, the key with the highest overall score will be marked as the primary peak key.

6. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 1, characterized in that, Output the keystroke detection result, including: If a main peak key or combination key exists, activate the single key discriminator. The single-key discriminator takes the pressure value of the pressure sensing process of the main peak key as input, performs pressure sensing in stages on the key pressing process, performs binary classification judgment under the extraction of contact morphology feature vector, and determines the key press judgment result, wherein the key press judgment result is either non-misclick or misclick.

7. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 6, characterized in that, The button pressing process is divided into three phases: pressing, holding, and releasing. The pressing phase is defined as the period from the point of contact with the key surface to the point of maximum pressure, the holding phase is defined as the stable range near the point of maximum pressure, and the lifting phase is defined as the period from the point of maximum pressure to the point where the fingertip leaves the key surface.

8. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 7, characterized in that, For each stage, contact morphology features are extracted to form a contact morphology fingerprint vector, and then assigned and judged according to each feature category. The feature categories include center of gravity trajectory, contact area change, pressure distribution pattern and peak features. The center of gravity trajectory is the two-dimensional displacement trajectory of the pressure center of gravity in the key surface. The contact area change is the expansion or contraction curve of the contact area over time. The pressure distribution pattern is the skewness and anisotropy of the pressure distribution. The peak features are the maximum pressure value and the slope of pressure rise and fall.

9. The method for identifying accidental key presses on a keyboard based on pressure sensing as described in claim 1, characterized in that, Based on the keystroke determination result, if it is determined to be a non-misclick, the key code of the non-misclick key is added to the input event queue. Based on the input event queue, key presses and character inputs at the operating system level are triggered; If the key press is determined to be accidental, the key event is discarded and the current trigger is marked as invalid, and no key message is sent to the host.

10. A keyboard key mis-touch recognition system based on pressure sensing, characterized in that, For implementing the pressure-sensitive keyboard key mis-touch recognition method according to any one of claims 1-9, the system comprises: The identification threshold deployment module uses short-term benchmark typing data as a basis to deploy the first identification threshold by assigning differentiated threshold coefficients to different areas of the keyboard. The pressure heat map acquisition module is used to collect keyboard pressure values ​​through real-time pressure sensing and map them to a two-dimensional keyboard coordinate space, and to determine the effective pressure heat map based on the first recognition threshold for hard boundary determination. The pattern determination result acquisition module is used to perform pattern determination based on keystroke pattern for the effective pressure heat map to obtain a first-order determination result, wherein the first-order determination result is a false touch or a non-false touch, and the non-false touch is marked with the main peak key position or combination key. If the first-order determination result is not a mis-touch, the keystroke determination result output module activates the single-key discriminator, which divides the keystroke process into the pressing stage, holding stage and lifting stage, extracts the contact pattern fingerprint vector and performs binary classification determination, and outputs the keystroke determination result. The keystroke determination result execution module reports the key code to trigger an input response if the keystroke determination result is not a mis-touch, and does not generate a system event if the keystroke determination result is a mis-touch.

Images