Motion-Pattern Recognition System Using a Wavelet-Neural Network
Wen-Ren Yang, Chau‐Shing Wang, Chien-Pu Chen · IEEE Transactions on Consumer Electronics · 2019
This paper presents a wavelet-neural network recognition system for personal fitness assistance and elderly daily activity monitoring applications. This discrete wavelet transform radial basis neural network (DWT-RBNN) analyzes vibration induced by human motions, and identifies motion status automatically. The 3-D vibration signals are measured by integrated accelerometer chip, and then DWT extracts vibration features. Local energy of extracted feature is calculated and used by RBNN. A multi-channel RBNN is designed and used for recognition. The computation burden is reduced because of the DWT pre-processing. From experiment results, RBNN shows successful recognition capability. This paper also presents flow diagram to determine engineering parameters for the present and future product developments.