Gait recognition based on Fourier descriptors and Canonical Time Warping

摘要

The Gait recognition includes the 3 main stages: In stage of moving target detection, this paper uses the background subtraction to extract gait silhouette. In stage of feature extraction, using Key frame technology and Fourier descriptors to extract the motion sequence silhouette feature. According to obtain feature data, we can use weighted Knearest neighbor (KNN) classifier to train gait database and recognize gait sequence with Canonical Time Warping (CTW).