Adaptive Filter Transform for Capacitive Touch Sensor Noise Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Capacitive touch screens face challenges in reliably detecting inputs due to manufacturing variations and noise interference, leading to potential errors in touch detection and tracking.
Innovation Solution
A method and device that perform regression analysis on sensor data to determine appropriate filter transforms, such as IIR, first-order, or second-order filters, based on coefficient comparisons with threshold values, to generate output matrices for accurate touch detection and tracking, while adapting to no-interaction scenarios by skipping regression analysis and using baseline measurements to reduce noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If regression analysis is performed on sensor data to determine filter transforms, then touch detection accuracy is improved, but computational resources and processing time are increased
Solution Approach 1:
The system dynamically selects the appropriate filter transform type (IIR, first-order, or second-order) based on regression analysis coefficients and threshold comparisons. This adaptive approach allows the system to use more complex processing only when necessary, while using simpler, faster processing for common cases, thus resolving the contradiction between accuracy and speed.
Solution Approach 2:
The system changes the filter transform parameters based on the regression analysis results. By comparing coefficients against threshold values, the system selects different filter orders and types, optimizing the balance between processing accuracy and computational efficiency for each specific sensor data scenario.
2Reliability
If filter transforms are applied to sensor data, then noise is reduced and detection reliability is improved, but processing complexity is increased
Solution Approach 1:
The system applies different filter transforms to different rows of sensor data based on local characteristics detected through regression analysis. Each row can have its own optimal filter type selected based on its specific coefficient patterns, allowing targeted noise reduction without unnecessarily complex processing of the entire sensor array.
Solution Approach 2:
The system performs regression analysis and filter selection only for rows where interactions are detected, skipping unnecessary processing for rows without interactions. This partial action approach reduces overall processing complexity while maintaining high reliability for the rows that do require filtering.
3Productivity
If baseline measurements are used for no-interaction scenarios, then computational resources are reduced, but detection precision may be compromised
Solution Approach 1:
The system uses feedback from regression analysis coefficients to determine when baseline measurements are appropriate. By continuously monitoring coefficient values against thresholds, the system can confidently switch between baseline-based processing (for efficiency) and full filter transform processing (for precision), ensuring that precision is maintained when needed while maximizing efficiency otherwise.
Data Source
AI summary
According to an embodiment, a regression analysis is performed on a subset of a dataset, where the subset of the dataset corresponds to inputs from a first row of a matrix of sensors at a time instant k. The regression analysis generates a set of coefficients. A filter transform, to be applied on the subset of the dataset, is determined based on a comparison between the set of coefficients and threshold values. The filter transform can be one of an infinite impulse response (IIR) filter transform, a first-order filter transform, or a second-order filter transform. Once the filter transform is determined, it is applied to the subset of the dataset to generate a first output matrix.