Amplitude Modulation and Convolutional Encoder Techinques for Gait Speed Classification

摘要

Human gait analysis can provide vital information especially with optical sensor to track and asses the changes in gait pattern. In this paper, the Kinect v2 based system is used for classifying a several walk speeds and Convolutional Encoder (CE) technique is validated. Positional skeletal data is collected from the lower body's joints and Amplitude Modulation (AM) is used to modify gait signal for extracting the gait features namely baseband frequency and modulation index. The obtained results show that 97.8% of the considered parameters are appropriate for classifying between three kinds of walk speeds.