Research And Implementation Of Biometrics Based On Face And Gait Characteristics

Social public security has been attracted more and more attention from the government and the society.Moreover,it has become a government affair that the government especially the public security political and legal institutions must do well.In order to grasp the real-time information in public places,video monitoring system is installed everywhere,especially in cities.With the rapid popularization of video monitoring,to quickly confirm the identity of personnel and realize intelligent warning,a rapid identification technology being independent of distance and non-cooperation of users is necessary.In this paper,multi-modal identification method is realized by combining face feature and gait feature.In the middle distance,Lacus-Kanade optical flow method was used to extract the gait optical flow image during human walking to characterize the dynamic characteristics of gait,and Procrustes shape analysis method was used to calculate head and shoulder shape of a gait.At the same time,the walking direction of a pedestrian is estimated according to the relation between the height of the silhouette at the beginning and the end of a gait cycle,and the perspective between the pedestrian and the camera is determined as the selection criteria of gait characteristics.This method overcomes the influence of the perspective on gait recognition.In case of closed distance,face features are used for identification,and sparse representation recognition strategy based on sub-blocks is proposed.According to certain proportion relationship,a face was blocked to four parts.Then,multi-scale image characteristics of each block was extracted.A random perception matrix is designed to map face features from high dimension space to low-dimensional subspace.The stochastic perception matrix has the property of data independence.This method improves the face recognition rate and satisfies the real-time requirement of the system.Finally,decision level fusion of gait recognition and face recognition results is carried out to solve the problem of identification difficulty in the case of remote distance.The experimental results have shown that the presented method can provide meaningful reference for the design of the biometric identification in the remote distance.