Motion Analysis and Recognition Using Principle Component Analysis and Angular Velocity

Abstract

Wearable sensor based and vision based are the two common methods used to analyze motion. These methods mainly used in medical and sports field where the study of body parts is crucial. In this study, 3-space sensor was used to collect the data of walking and jogging motions of subjects. Angular velocity was proposed as the component to recognize human motions. The data were processed by Principle of Component Analysis (PCA) and recognized by MATLAB functions. PCA was used to reduce and categorize the data into the respective principle components for recognition. Vectors were generated in the system to re-present the motions of the complete recorded activities. The proposed motion recognition system achieved more than 85 % of accuracy.