REAL-TIME ACCELEROMETER SIGNAL PROCESSING OF END POINT DETECTION AND FEATURE EXTRACTION FOR MOTION DETECTION
JG Lim, Yi Sohn, Dong‐Soo Kwon · 2007
Abstract: A simple signal processing algorithm is proposed for a novel input device developed for the elderly. The device is conceptually based on a pen, since it is one of the most familiar communication tools for the elderly and is closely tied to Hand Writing Recognition. We propose a new algorithm to execute End Point Detection and feature extraction for pattern recognition simultaneously. This algorithm is based on the fact that the drift error during acceleration can be corrected using the Zero Velocity Compensation, which is sensitive to the performance of EPD. The performance of the proposed algorithm is compared to that of the conventional approaches and its recognition rate is found to be superior to that of a previous research. Copyright © 2007 IFAC