Infrared sensor-based pressure touch detection method for armrest screen

Abstract

This invention discloses a pressure touch detection method for armrest screens based on infrared sensors. This invention relates to the field of touch detection technology and solves the technical problems of poor environmental adaptability and susceptibility to interference-induced false touches. By detecting acceleration before startup and calibrating only when the device is stationary, this invention eliminates vibration interference during transportation or installation. It utilizes a reference optical path to calculate the ambient light attenuation coefficient in real time, automatically offsetting signal drift caused by changes in sunlight and lighting, ensuring a high signal-to-noise ratio under both strong and weak light conditions. Combining the background noise standard deviation with dynamic threshold calculation, the system can adapt to noise levels in different environments, reducing false touch and missed touch rates. A centroid queue is constructed, effectively utilizing the temporal correlation of historical data. A dynamic anomaly removal logic based on region area is introduced, intelligently identifying and discarding outliers caused by transient interference. The final output of effective touch coordinates has extremely high stability, improving the finesse of operation.

Description

Technical Field

[0001] This invention relates to the field of touch detection technology, specifically to a method for detecting pressure touch on an armrest screen based on an infrared sensor. Background Technology

[0002] Infrared touchscreens are widely used in public information terminals, industrial control panels, and in-vehicle entertainment systems due to their advantages such as good light transmittance, resistance to surface contamination, and support for glove operation. A typical infrared touchscreen uses an array of infrared emitters and receivers arranged around the edge of the screen to form a crisscrossing infrared light grid. When a finger or object blocks the light, the light intensity at the receiver changes, and the system calculates the touch coordinates based on the change in position.

[0003] According to patent publication number CN117453066A, a pressure-sensitive wake-up armrest screen device is disclosed, including a housing, a touch screen display, a sensor module, a circuit mechanism, and a control unit. The sensor module measures and monitors the force applied to the touch screen display. When the screen pressure detected by the sensor module is less than 1N, the sensor module does not provide feedback but continues to collect data in real time. At this time, the circuit mechanism and force sensor enter a low-power mode. When the screen pressure detected by the sensor module is greater than or equal to 1N, the sensor module sends a wake-up signal to the circuit mechanism, causing the circuit mechanism and force sensor to enter normal operating mode and restore normal operating current. Power supply to the on-chip system chip and touch screen display is also restored, and the entire screen starts up and resumes display and touch functions. This invention reduces the static current of the armrest screen without adding physical buttons, helping to reduce energy loss and supporting the advocacy and practice of green and low-carbon living.

[0004] Traditional infrared touch systems typically use a fixed sensitivity threshold for signal detection. When the ambient light intensity changes drastically or there is slight vibration, the photoelectric signal at the receiving end will fluctuate, easily causing false triggers. Furthermore, due to the thermal noise of the infrared sensor itself, circuit noise, and external interference, single-frame data exhibits random fluctuations, resulting in noticeable jitter of the mouse pointer or cursor, severely impacting the user experience. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a pressure touch detection method for armrest screens based on infrared sensors, which solves the problems of poor environmental adaptability and susceptibility to interference-induced accidental touches.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting pressure touch on an armrest screen based on an infrared sensor, which specifically includes the following steps: Step 1: When the device is powered on, it is forced into no-load mode. After the vibration stabilizes, a full array scan is performed. The original analog voltage values ​​of each receiving tube are recorded and converted into light intensity values. An ambient light attenuation coefficient is introduced to correct the light intensity values, and the corrected relative light intensity values ​​are obtained. A dynamic reference library is established simultaneously. Step 2: Scan at a preset frequency in time division, compare the currently collected relative light intensity value with the corresponding benchmark value in the dynamic benchmark library, calculate the light intensity change, calculate the dynamic threshold by combining the current ambient light decay coefficient and background noise standard deviation, filter the set of pixels with light intensity change greater than the dynamic threshold, calculate the confidence level, and determine the effective touch area. Step 3: Calculate the centroid coordinates based on the coordinates of each pixel within the effective touch area and their relative light intensity changes. Construct a queue to store the centroid coordinates. Calculate the maximum displacement of each point in the queue relative to the mean and filter the effective points. Take a weighted average of the remaining effective points to obtain stable touch coordinates. Obtain the touch pressure, touch area, and dwell time. Compare each with its respective threshold range to obtain the feature score. Sum the scores to obtain a weighted total score. If the weighted total score is greater than the scoring threshold, it is determined to be a touch event. Step 4: Obtain the original touch flow based on touch events, extract geometric topological features, temporal dynamic features, and physical interaction features, integrate them into the feature vector at the current moment, read the current system context state, extract the corresponding weight vector from the preset dynamic weight library, and generate the scene adaptation feature vector by weighted summation of the feature vector and the weight vector. Calculate the similarity between the scene adaptation feature vector and the predefined standard gesture template library, filter out templates with similarity greater than the preset value, and if multiple templates exist, select the gesture template with the highest comprehensive score according to the secondary sorting of the dynamic weight library as the recognition result and execute the corresponding instruction.

[0007] As a further aspect of the present invention, the specific method for converting the original analog voltage value into a light intensity value is as follows: Read the raw analog voltage value and dark voltage reference value According to the formula The differential voltage was calculated. By calibrating with a standard laboratory light source, a series of voltage values ​​corresponding to known light intensities were recorded, and a mapping function was constructed. The light intensity value I is calculated based on this mapping function.

[0008] As a further aspect of the present invention, the method of introducing an ambient light attenuation coefficient to correct the light intensity value is as follows: Record the initial light intensity of the reference optical path. Real-time monitoring of the light intensity of the current reference optical path According to the formula Calculate the ambient light decay coefficient K; all actual light intensity values ​​collected need to be multiplied by K. After normalization, the corrected relative light intensity value is obtained. .

[0009] As a further aspect of the present invention, the method for calculating the dynamic threshold by combining the current ambient light attenuation coefficient and the standard deviation of background noise is as follows: The system reads the ambient light attenuation coefficient K and the standard deviation of the background noise of the current frame. And according to the formula Calculate the dynamic threshold ,in This is the factory-calibrated baseline sensitivity threshold. This is the environmental compensation coefficient. This is the noise sensitivity coefficient.

[0010] As a further aspect of the present invention, the specific method for determining the effective touch area is as follows: The system collects all that meet the requirements. > The set S of pixels whose light intensity change is greater than the dynamic threshold is traversed by the 8-neighborhood search algorithm to determine whether each pixel has adjacent pixels in the up, down, left, right and diagonal directions. If there are at least two adjacent pixels, it is marked as a connected region. Calculate the geometric features of the connected region according to the formula. The normalized area was calculated. ,in Represents the area of ​​the connected region. This represents the maximum reference area, and according to the formula... The circularity feature C is calculated, where Area represents the total number of pixels in the connected region and Perimeter represents the perimeter. According to the calculation formula The confidence score of the connected components is obtained, and and These are the corresponding weight coefficients; finally, the confidence score (Score) is compared with the confidence threshold (Sth). If Score > Sth, it is determined to be a valid touch area.

[0011] As a further aspect of the present invention, the stable touch coordinates are obtained as follows: Based on the obtained effective touch area The coordinates (xi, yi) of all pixels in the area are collected simultaneously, and the relative light intensity change corresponding to each pixel is extracted. According to the formula Calculate the centroid to obtain its coordinates. ; Construct a queue of length N based on the obtained centroid coordinates to store the original centroid coordinates of the most recent N frames. Calculate the maximum displacement of each point in the queue. ,in and This represents the queue mean, if the maximum displacement... If the area exceeds n times the current region's area, remove the outlier. Repeat this process to remove all outliers. Then, calculate a weighted average for the remaining valid points. Finally, stable touch coordinates are obtained, which are recorded as valid touch coordinates.

[0012] As a further aspect of the present invention, the method for determining the touch event is as follows: For the touch area, obtain the area of ​​the effective touch area and the normal touch area. If the area of ​​the effective touch area is larger than the normal touch area, it is judged as a false touch; otherwise, it is marked as an effective touch area. Regarding dwell time, the touch time is obtained by measuring the time from when the touch appears to when it disappears. This time is then compared with a time threshold range: if the touch time is less than the minimum time threshold or greater than the maximum time threshold, it is determined to be a false touch; otherwise, it is marked as a valid touch. Next, the results of the comprehensive touch feature recognition are weighted and summed, with each feature receiving a score of 0 or 1; 0 points for a mis-touch and 1 point for a normal touch. Simultaneously, different touch features are assigned corresponding weights, which are then combined with the calculation formula. The weighted total score S is calculated, where , The scores represent touch pressure, touch area, and dwell time, respectively. , and are the corresponding weights. The weighted total score S is compared with the scoring threshold. If it is greater than the scoring threshold, it is determined to be a touch event.

[0013] As a further aspect of the present invention, the feature vector at the current moment is obtained in the following way: The raw touch flow Et is acquired in real time via a touch sensor, where Et Includes pressure Parameters such as contact area are extracted; then, the geometric topological features, temporal dynamic features, and physical interaction features of each frame t are extracted, with the geometric topological features being the curvature of the current trajectory. The temporal dynamics characteristic is instantaneous velocity. The physical interaction characteristic is the mean pressure. The above features are then integrated into the feature vector at the current time. .

[0014] As a further aspect of the present invention, the method for generating the scene adaptation feature vector is as follows: Read the current system context state, which includes application type, focused control type, and UI level. Based on this state, extract the corresponding weight vector from the preset dynamic weight library. Then, the feature vector and weight vector at the current moment are summed in a weighted manner to generate the scene-adaptive feature vector.

[0015] This invention provides a method for detecting pressure touch on an armrest screen based on an infrared sensor. Compared with existing technologies, it has the following advantages: This invention eliminates vibration interference during transportation or installation by detecting acceleration before startup and calibrating only when the device is stationary. It uses a reference optical path to calculate the ambient light attenuation coefficient in real time, automatically offsetting signal drift caused by changes in sunlight and lighting, ensuring a high signal-to-noise ratio under both strong and weak light conditions. It also calculates a dynamic threshold by combining the background noise standard deviation, enabling the system to adapt to different noise levels in different environments, thus reducing false touch rate and missed touch rate.

[0016] This invention effectively utilizes the temporal correlation of historical data by constructing a centroid queue and introduces dynamic anomaly removal logic based on region area. It can intelligently identify and discard outliers caused by instantaneous interference, without causing delays due to excessive smoothing like traditional mean filtering. The final output of effective touch coordinates has extremely high stability, solves the cursor jitter problem, and improves the finesse of operation. Attached Figure Description

[0017] Figure 1 This is a flowchart of the armrest screen pressure touch detection method of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1 This application provides a method for detecting pressure touch on an armrest screen based on an infrared sensor, which specifically includes the following steps: Step 1: When the device is powered on, it is forced into no-load mode. No-load mode means that there are no objects obstructing the screen surface and the armrest screen is in a stationary state. During operation, if the system detects no touch signal input for a continuous period of T (the specific value of T is set by the operator) or receives a calibration command initiated by the user, it will automatically trigger a full array scan. Before scanning, the acceleration sensor data must be read. If the device is detected to be in a state of severe vibration, the scan will be suspended and executed again after the vibration stabilizes. Here, stable vibration means acceleration threshold <0.5g. Then, the infrared emitting diodes are controlled to light up sequentially in a random scanning order, while the original analog voltage value of each receiving diode at the current moment is recorded. All the collected original analog voltage values ​​are then processed. Convert to light intensity value I. The specific conversion method is as follows: read the original analog voltage value. and dark voltage reference value According to the formula The differential voltage was calculated. Next, using a standard laboratory light source for calibration, a series of voltage values ​​corresponding to known light intensities were recorded, and a mapping function was constructed. (The specific expression is:) in Indicates the calibrated light intensity range. Indicates the rated voltage range. (representing the normalization scaling factor), and the corresponding light intensity value I is calculated based on this mapping function; Simultaneously, an ambient light attenuation coefficient is introduced to correct the obtained light intensity value I. The specific correction method is as follows: record the initial light intensity of the reference optical path. Real-time monitoring of the light intensity of the current reference optical path According to the formula Calculate the ambient light decay coefficient K; all actual light intensity values ​​collected need to be multiplied by K. After normalization, the corrected relative light intensity value is obtained. A dynamic benchmark library is established simultaneously.

[0020] Step 2: Entering the real-time touch detection stage, the system continuously performs time-division scanning of the infrared emitting diode array at a preset frequency. Within each scanning cycle, the system sequentially drives each infrared emitting diode to light up in a preset order, while the corresponding receiving diodes collect light intensity signals in real time and convert them into relative light intensity values. Subsequently, the system will display the currently collected relative light intensity value. The light intensity change is calculated by comparing it with the reference light intensity value at the corresponding location in the dynamic reference library. Simultaneously, the system reads the ambient light attenuation coefficient K and the standard deviation of the background noise of the current frame. The standard deviation of background noise The fluctuation values ​​of the most recent M frames are obtained by using a sliding window, and then calculated according to the formula. Calculate the dynamic threshold ,in This is the factory-calibrated baseline sensitivity threshold. This is the environmental compensation coefficient. This is the noise sensitivity coefficient.

[0021] Next, based on the obtained dynamic threshold and changes in light intensity The system collects all that meet the requirements. > The set S of pixels whose light intensity change is greater than the dynamic threshold is identified, and this set is then evaluated: if the number of pixels in S is less than 2, they are considered random noise and discarded; if the number of pixels in S is greater than or equal to 2, it is necessary to check whether these pixels form a connected region. The specific evaluation method is as follows: The system iterates through the pixels in set S using an 8-neighborhood search algorithm, determining whether each pixel has adjacent pixels in the top, bottom, left, right, and diagonal directions. If at least two adjacent pixels exist, the pixel is marked as a connected region. Subsequently, the system calculates the geometric features of this connected region according to the formula... The normalized area was calculated. ,in Represents the area of ​​the connected region. This represents the maximum reference area, and according to the formula... The circularity feature C is calculated, where Area represents the total number of pixels in the connected region and Perimeter represents the perimeter. Next, substitute the calculated parameters into the calculation formula. The confidence score of the connected components is obtained, and and These are the corresponding weight coefficients; finally, the confidence score (Score) is compared with the confidence threshold (Sth). If Score > Sth, it is determined to be a valid touch area; otherwise, it is determined to be a false touch and the data in that area is cleared.

[0022] Step 3: Based on the obtained valid touch area The coordinates (xi, yi) of all pixels in the area are collected simultaneously, and the relative light intensity change corresponding to each pixel is extracted. Then, according to the formula... Calculate the centroid to obtain its coordinates. Construct a queue of length N based on the obtained centroid coordinates to store the original centroid coordinates of the most recent N frames. Calculate the maximum displacement of each point in the queue. ,in and This represents the queue mean. If the maximum displacement... If the area exceeds n times the current region's area, remove the outlier. Repeat this process to remove all outliers. Then, calculate a weighted average for the remaining valid points. Finally, stable touch coordinates are obtained, which are recorded as valid touch coordinates; Based on the obtained valid touch coordinates, touch features are acquired, including touch pressure, touch area, and dwell time, and each touch feature is identified and judged separately. For touch pressure, the deformation of the structural pressure spring is detected by an infrared sensor, and the corresponding real-time pressure is obtained through software calibration. This real-time pressure is compared with a pressure threshold range (e.g., [Pmin, Pmax]): if the real-time pressure is greater than the maximum pressure threshold or less than the minimum pressure threshold, it is determined to be a false touch; otherwise, it is marked as a valid touch. Regarding touch area, obtain the area of ​​the effective touch area and the normal touch area. For example, the normal touch area of ​​a finger is usually 30-120mm². 2 Between these, the palm area is much larger than the normal fingertip and the shape is more oval or scattered. If the area is larger than the normal touch area, it is judged as a false touch; otherwise, it is marked as a valid touch area. Special cases such as water droplets or stains are not considered here. Regarding dwell time, the touch time is obtained by measuring the time from when the touch appears to when it disappears. This time is then compared with a time threshold range: if the touch time is less than the minimum time threshold or greater than the maximum time threshold, it is determined to be a false touch; otherwise, it is marked as a valid touch. Next, the results of the comprehensive touch feature recognition are weighted and summed. Each feature is scored either 0 or 1, with 0 points for a mis-touch and 1 point for a normal touch; at the same time, different touch features are assigned corresponding weights, combined with the calculation formula. The weighted total score S is calculated, where , The scores represent touch pressure, touch area, and dwell time, respectively. , and are the corresponding weights. The weighted total score S is compared with the scoring threshold. If it is greater than the scoring threshold, it is determined to be a touch event.

[0023] Step 4: Based on the determined touch event, acquire the original touch stream Et in real time through the touch sensor, where Et Includes pressure Parameters such as contact area are extracted; then, the geometric topological features, temporal dynamic features, and physical interaction features of each frame t are extracted, with the geometric topological features being the curvature of the current trajectory. The temporal dynamics characteristic is instantaneous velocity. The physical interaction characteristic is the mean pressure. The above features are then integrated into the feature vector at the current time. ; Next, the current system context state is read, including the application type, the type of focused control, and the UI level. Based on this state, the corresponding weight vector is extracted from the preset dynamic weight library. If in text editing mode, increase the weight of position overlap and decrease the weight of trajectory length; if in image browsing mode, increase the weight of zoom and decrease the weight of directionality; then, the feature vector at the current moment is summed with the weight vector to generate a scene-adaptive feature vector. Simultaneously, the scene-adapted feature vector is compared with a predefined standard gesture template library in real time for similarity calculation. The cosine similarity formula is used for similarity calculation to select gesture templates with real-time similarity greater than the preset similarity value. If multiple templates meet the conditions, the similarity is sorted a second time according to the weight of each feature dimension in the dynamic weight library, and the gesture template with the highest comprehensive score is selected as the recognition result. If the real-time similarity is lower than the preset similarity value, it is determined to be an unrecognized gesture, and the system does not perform any operation.

[0024] Some of the data in the above formulas are numerical calculations with dimensions removed, and the contents not described in detail in this specification are all prior art known to those skilled in the art.

[0025] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for detecting pressure touch on an armrest screen based on an infrared sensor, characterized in that, The method specifically includes the following steps: Step 1: When the device is powered on, it is forced into no-load mode. After the vibration stabilizes, a full array scan is performed. The original analog voltage values ​​of each receiving tube are recorded and converted into light intensity values. An ambient light attenuation coefficient is introduced to correct the light intensity values, and the corrected relative light intensity values ​​are obtained. A dynamic reference library is established simultaneously. Step 2: Scan at a preset frequency in time division, compare the currently collected relative light intensity value with the corresponding benchmark value in the dynamic benchmark library, calculate the light intensity change, calculate the dynamic threshold by combining the current ambient light decay coefficient and background noise standard deviation, filter the set of pixels with light intensity change greater than the dynamic threshold, calculate the confidence level, and determine the effective touch area. Step 3: Calculate the centroid coordinates based on the coordinates of each pixel within the effective touch area and their relative light intensity changes. Construct a queue to store the centroid coordinates. Calculate the maximum displacement of each point in the queue relative to the mean and filter the effective points. Take a weighted average of the remaining effective points to obtain stable touch coordinates. Obtain the touch pressure, touch area, and dwell time. Compare each with its respective threshold range to obtain the feature score. Sum the scores to obtain a weighted total score. If the weighted total score is greater than the scoring threshold, it is determined to be a touch event. Step 4: Obtain the original touch flow based on touch events, extract geometric topological features, temporal dynamic features, and physical interaction features, integrate them into the feature vector at the current moment, read the current system context state, extract the corresponding weight vector from the preset dynamic weight library, and generate the scene adaptation feature vector by weighted summation of the feature vector and the weight vector. Calculate the similarity between the scene adaptation feature vector and the predefined standard gesture template library, filter out templates with similarity greater than the preset value, and if multiple templates exist, select the gesture template with the highest comprehensive score according to the secondary sorting of the dynamic weight library as the recognition result and execute the corresponding instruction.

2. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The specific method for converting the original analog voltage value into a light intensity value is as follows: Read the raw analog voltage value and dark voltage reference value According to the formula The differential voltage was calculated. By calibrating with a standard laboratory light source, a series of voltage values ​​corresponding to known light intensities were recorded, and a mapping function was constructed. The light intensity value I is calculated based on this mapping function.

3. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The method of correcting the light intensity value by introducing an ambient light attenuation coefficient is as follows: Record the initial light intensity of the reference optical path. Real-time monitoring of the light intensity of the current reference optical path According to the formula Calculate the ambient light decay coefficient K; all actual light intensity values ​​collected need to be multiplied by K. After normalization, the corrected relative light intensity value is obtained. .

4. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The method for calculating the dynamic threshold by combining the current ambient light attenuation coefficient and the standard deviation of background noise is as follows: The system reads the ambient light attenuation coefficient K and the standard deviation of the background noise of the current frame. And according to the formula Calculate the dynamic threshold ,in This is the factory-calibrated baseline sensitivity threshold. This is the environmental compensation coefficient. This is the noise sensitivity coefficient.

5. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The specific method for determining the effective touch area is as follows: The system collects all that meet the requirements. > The set S of pixels whose light intensity change is greater than the dynamic threshold is traversed by the 8-neighborhood search algorithm to determine whether each pixel has adjacent pixels in the up, down, left, right and diagonal directions. If there are at least two adjacent pixels, it is marked as a connected region. Calculate the geometric features of the connected region according to the formula. The normalized area was calculated. ,in This represents the area of ​​the connected region. This represents the maximum reference area, and according to the formula... The circularity feature C is calculated, where Area represents the total number of pixels in the connected region and Perimeter represents the perimeter. According to the calculation formula The confidence score of the connected components is obtained, and and These are the corresponding weight coefficients; finally, the confidence score (Score) is compared with the confidence threshold (Sth). If Score > Sth, it is determined to be a valid touch area.

6. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The stable touch coordinates are obtained as follows: Based on the obtained effective touch area The coordinates (xi, yi) of all pixels in the area are collected simultaneously, and the relative light intensity change corresponding to each pixel is extracted. According to the formula Calculate the centroid to obtain its coordinates. ; Construct a queue of length N based on the obtained centroid coordinates to store the original centroid coordinates of the most recent N frames. Calculate the maximum displacement of each point in the queue. ,in and This represents the queue mean, if the maximum displacement... If the area exceeds n times the current region's area, remove the outlier. Repeat this process to remove all outliers. Then, calculate a weighted average for the remaining valid points. Finally, stable touch coordinates are obtained, which are recorded as valid touch coordinates.

7. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The method for determining the touch event is as follows: For the touch area, obtain the area of ​​the effective touch area and the normal touch area. If the area of ​​the effective touch area is larger than the normal touch area, it is judged as a false touch; otherwise, it is marked as an effective touch area. Regarding dwell time, the touch time is obtained by measuring the time from when the touch appears to when it disappears. This time is then compared with a time threshold range: if the touch time is less than the minimum time threshold or greater than the maximum time threshold, it is determined to be a false touch; otherwise, it is marked as a valid touch. Next, the results of the comprehensive touch feature recognition are weighted and summed, with each feature receiving a score of 0 or 1; 0 points for a mis-touch and 1 point for a normal touch. Simultaneously, different touch features are assigned corresponding weights, which are then combined with the calculation formula. The weighted total score S is calculated, where , The scores represent touch pressure, touch area, and dwell time, respectively. , and are the corresponding weights. The weighted total score S is compared with the scoring threshold. If it is greater than the scoring threshold, it is determined to be a touch event.

8. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The feature vector at the current moment is obtained in the following way: The raw touch flow Et is acquired in real time via a touch sensor, where Et Includes pressure Parameters such as contact area; Subsequently, the geometric topological features, temporal dynamic features, and physical interaction features of each frame t are extracted. The geometric topological features are the curvature of the current trajectory. The temporal dynamics characteristic is instantaneous velocity. The physical interaction characteristic is the mean pressure. The above features are then integrated into the feature vector at the current time. .

9. The armrest screen pressure touch detection method based on an infrared sensor according to claim 1, characterized in that, The method for generating the scene adaptation feature vector is as follows: Read the current system context state, which includes application type, focused control type, and UI level. Based on this state, extract the corresponding weight vector from the preset dynamic weight library. Then, the feature vector and weight vector at the current moment are summed in a weighted manner to generate the scene-adaptive feature vector.

Images