Based On The Video Gait Research

Gait recognition is a new technology to identify and recognize persons according their working styles, i.e. gaits. Owing to the development of computer science and to the updating performance of computer, as well as the complex algorithm proposed, gait recognition has become a hot topic in the area of machine learning. Compared to other biometric feature, gait is of the characteristic of no-touch ,easy access, un- intrusion, difficult to disguise and hide. In the area of surveillance, human-machine interaction, medical diagnosis, etc, the study of gait recognition is of extreme significance and has extensive application prospect. Gait recognition is including four main parts: background modeling, image preprocessing, feature extraction and recognition, among them, feature extraction is the core technology. This paper elaborate the method of feature extraction andThis paper focus on the study of gait , the main work in this paper consist of the following parts:i) The research significance of gait recognition is first introduced, then analyze the common background modeling methods and object detection algorithm is proposed.ii) The median algorithm is adopted to create background, background elimination is used to detect motion object. Mathematical morphology method and region denoting are applied to process binary image.iii) Then a new method to compute gaits periodic is proposed.iv) Using contour tracing method to extract the contour of a working person, a eigen vector is acquired based on the distance from the point at the contour and to the centroid.v) Jaccobi method is utilized to compute eigenvalues and eigenvectors, then a eigenspace is constructed according to each gait mode.