Adaptive Kalman Filter: Noise Reduction in Diagonal Drawings on Stylus/Pen Touchscreens for Enhanced Precision

Abstract

The adaptive Kalman filtering algorithm, designed to accommodate the dynamic nature of the system, provides an adaptive estimation of the state by incorporating both process and measurement noise considerations, thereby effectively reducing the noise and preserving the integrity of diagonal line drawings. The iterative prediction and update process employed by the algorithm aids in achieving smoother and more accurate position estimations. To assess the efficacy of the adaptive Kalman filtering approach, a comparative analysis was performed against a multistage filter. This filter employed a sequence of median filters with progressively increasing window sizes to eliminate outliers and artifacts while retaining the intricate details of the drawings. A comprehensive evaluation was performed via a detailed comparison of noise reduction performance and preservation of details between the two techniques. The experimental findings unequivocally established the superiority of the adaptive Kalman filtering approach in noise reduction and accuracy enhancement of the recorded positions. The proposed algorithm surpassed the multistage filter, demonstrating superior noise reduction capabilities while maintaining the desired level of detail in diagonal line drawings. The findings are expected to contribute to the advancement of state estimation techniques in dynamic systems, with a focus on augmenting accuracy and detail preservation. KEYWORDS dynamic system, impulse noise, measurement noise, multistage filter, outlier removal, position estimation