Application of КРСА with different kernels for human gait assessment

Abstract

The evaluation of Kernel Principal Component Analysis (KPCA) based on different kernels in human gain assessment has been made. Three types of kernel have been chosen: linear, polynomial and RBF. The normalcy index was computed to evaluate usefulness kernels in human gait assessing. The test group consists of 45 persons (156 strides) with normal or pathological (Cerebral Palsy, Spina Biftda, Anterior Cruciate Ligament and Gonarthrosis) gait.