Research On Technology Of Human Identification Based On Gait

Gait recognition is a new kind of biometric features recognition technology. Gait can be used to recognize people by inspecting their walking manners regardless of their clothes. Particularly, it can assist to effectively recognize people at a distance. Moreover, gait is hard to be disguised or concealed. So, in recent years, it has become a research hotspot in the field of computer vision and pattern recognition, and has a broad application prospect in many domains, such as intelligent surveillance, medical diagnosis etc.This thesis research deeply on technologies of human identification based on gait, mainly including three parts:gait detection and preprocessing, feature extraction and classification. Gait feature extraction has been the focus of gait recognition, and is also a key part in this paper.Firstly, much research work on the method of gait detection and background modeling will be done in the paper. The background subtraction is used in motion detection; Gaussian model method is used in background modeling, then binary silhouette image is processed by Morphology and normalized operation, by the analysis of gait cycle, the key frame extraction algorithm is proposed to effectively remove redundant frames. Secondly, deep study on the method of gait feature extraction will be done in the paper. The theory of gait energy image is elaborated and probed that how to select effective feature and select what features, the feature selection mask method is proposed to improve gait energy image. By the research work on the theory of PCA and SVM, the feature extraction algorithm of this paper is proposed. Finally, on the basis of the theory, a simple and easy to operate gait recognition system is accomplished. Experiment on gait database of CASIA is used to analysis and verify the key technologies put forward in the paper. Experimental results show that the proposed algorithm achieves higher recognition rate.