Movement Pattern Recognition of Weight Lifter Based on Ground Reaction Force

Automatic movement recognition is a crucial part to the development of weight lifter training and evaluating system which uses the kinematic data from video-analyzing and dynamic information for diagnosing weight lifter's performance. Previous works focused mainly on video processing (kinematic data) for analyzing athlete's performance, which needs a tremendous computation. In this paper a novel approach to the problem was investigated, using the vertical component of an athlete's ground reaction force (GRF). Typical movement phases of a weight lifter are decomposed and recognized automatically in terms of the GRF signal measured by a force platform. Support vector machine (SVM) multi-classifier for the recognition of weight lifter's movement phases were implemented and the model parameters were determined by art and science. The classification results demonstrate the validity of SVM-based method