Research On Gait Recognition Method Based On Contour Feature And Multifractal Analysis

Gait Recognition,recognizing different person by the way of its walking,which has some characteristics such as non-contact,low demand in image quality and difficult disguise,is the most potentially biological recognition technology by the way of one's walking.Recently,the research of gait recognition is a hot topic in the field of computer vision,entrance guard system and medical diagnose,which provides with extensive realistic significance and wide applying foreground.The key steps of gait recognition are the gait feature,which is extracted to describe the motion of a cycle gait,and the corresponding classifier,which is used to accomplish eventual recognition.Combined with the accurate description of the detail information of gait frequency by Multifractal spectrum and the analysis of GEI(Gait Energy Image)and AFDEI(Accumulated Frame Difference Energy Image),which is constructed by the analysis the characteristics of gait sequence,this paper proposed a method of gait recognition using silhouette and multifractal analysis(GRSMA).The main work of this thesis includes:1.Made a summary for existing well-known gait recognition algorithms.They can be classified into as follows:model based methods and silhouette based methods which is divided into the method based on energy image and the method based on silhouette geometry feature.2.Got the AFDEI and DEI.Firstly,binary silhouette image is gained by preprocessing and the cycle of one's gait is determined by high-width ratio of the gait silhouette and then the silhouettes of one-period subsequence are extracted.Secondly,AFDEI is gained by the frame difference energy images,which are calculated by combining forward frame differences image with backward frame difference images,using weighted average method.Finally,GEI is obtained using weighted-average.3.Extracted the gait feature matrix based on the theory of multi fractal.Analyzed GEI and AFDEI in the view of texture analysis,and then obtained the multifractal spectrum of GEI and AFDEI respectively.The gait feature matrix was extracted after the analysis of the multifractal spectrum severally4.Obtained the experiment result by experimenting on the CASIA_B database.The final recognition results are obtained by fusing the gait feature matrixes of GEI and AFDEI.Compared with the algorithm GEI+W(2D)2PCA,GEI+(2D)2PCA and SFDEI+HMM,the experiment result showed that the algorithm proposed in thesis is better.