Application of Hybrid Multi-resolution Wavelet Decomposition Method in Detecting Human Walking Gait Events

Abstract

Identifying walking gait events is important in gait analysis. In particular, heel-strike and toe-off are commonly used to define the stance phase and swing phase in normal human walking gait cycle. They are used to segment a stream of human motion data into discrete and meaningful sections that can be analyzed and compared with available literatures. This paper proposes multi-resolution wavelet decomposition to reveal relevant information. Subsequently, proposed method differentiates the signal twice to identify the heel-strike and toe-off events. With this information, various temporal gait parameters can be easily estimated, such as the duration of swing phase and stance phase, and the duration of initial double support and terminal double support. Experimental results on the temporal parameters are comparable to the available benchmark data with minimal discrepancies due to the anthropometric properties of the subjects and inconsistent walking speed.