Research On Gait Recognition Method Based On Sequence Silhouette Feature

With the development of the computer vision technology,gait recognition technology has gradually become a hot spot in the field of identification.Gait recognition is based on person's walking style to identify individual,which is one of the biology identification technologies.Compared with other biology identification technologies,gait recognition has non-contact,long distance,difficult to disguise,low effect by external environment and so on.These advantages make it become a irreplaceable biology identification technology.So this paper studies the gait recognition technology in depth.Firstly,preconditioning technique of gait images is researched.Gait images are extracted from the video.Because the background of video is simple,moving object is extracted using background subtraction and object images are binarization.Images of noise are processed using morphological methods.As the size of object images is not uniform,the images are normalized,obtaining normalized gait silhouette images.Secondly,aiming at the problem of low recognition rate of single silhouette feature,gait recognition based on silhouette feature is proposed.The gait region area,the critical distance and the centroid to outermost silhouette distance of the lower part are fused.The recognition effect is remarkable.Finally,in order to improve the gait recognition rate and reflect the dynamic process of gait,gait recognition based on sequence silhouette variation feature.Sequence gait critical distance,sequence gait change and sequence maximum width feature are extracted.The recognition rate of sequence silhouette variation feature of the single and the fusion are analyzed,then compared with the recognition rate of the single silhouette feature.The gait region area and the sequence gait critical distance feature are fused,that is gait recognition based on sequence silhouette variation feature.A good recognition effect is obtained by this method in the condition of carrying a coat or a bag.