Recognition of Human Movement Based on Fourier Descriptors

A gait recognition method based on Fourier descriptors (FD) is put forward. First of all, by analyzing the gait cycle, gait sequence of the key frame could be subtracted from the image sequence, then we can get movement of the contours of human body through background abatement and use Fourier descriptors to handle key frames profile and compress the data. At last, we could get template matching, which can be classified and recognized by nearest neighbor law. The above method is experimented on the CASIA gait database. Experimental results demonstrate that this method has good recognition performance.