A Gait Recognition Method Based On Integrated Gabor Feature

With the status of machine vision in artificial intelligence field rising,gait recognition as one of biometric technology in this field has become gradually expanded.Gait technology is widely used in security protection,3D film,medicine and other fields.Gait varies in different ways,just like the various natural behaviors of people walking.Therefore,gait imitation is not easy to realize due to the lack of ability of identifying distance limitation,contact or other advantages,and exploring the physiological and psychological changes become the current mainstream research of the biometric techniques.In this thesis,a gait recognition method based on integrated Gabor feature is proposed.A comprehensive performance evaluation of the proposed method is carried out by CASIA Dataset B gait database provided by CAS Institute.The gait recognition is divided into three main processes:First process is gait detection.The external environment,shooting angle and shooting conditions have influences on gait target contour.Aimed to testify the above statement,a shadow elimination method based on HSV color model is used to extract the gait target contour graph effectively.Second process is gait feature extraction.According to the gait cycle image sequence,the dynamic region of the gait energy graph is extracted to be convoluted with the Gabor feature real part to obtain the gait feature graph.Aiming at identifying/proving the drawbacks of the traditional Gabor feature extraction,a Gabor integrated feature idea is proposed.According to the change of features,multi-angle and multi-scale feature graphs are integrated by mean fusion and differential binary coding.The last process is recognition.The 26 feature maps are aligned according to their recognition weights,four of which are selected as the final eigenvectors.The K nearest neighbor algorithm is used to classify them.The experimental results show that the gait recognition method based on the integrated Gabor feature is used to separate and express the gait features effectively,and the data dimension is reduced,so that the data is more compact and the gait is comparatively classified in a more accurate way.