Research On Gait Recognition Method Based On PHOG

Gait recognition is a special biometric recognition technology,which can achieve long-distance identification with low requirements on video quality,and is not easy to be detected and hidden.The application prospect in the field of intelligent monitoring is very broad.Therefore,the related technology research of gait recognition has been favored by many scholars in the field of computer vision,and the national government departments are paying more and more attention to the research and application of this technology.In the gait recognition process,how to weaken the influence of noise,improve the recognition rate,reduce the amount of calculation,and enhance the robustness of gait recognition,is always the goal of researchers in this field.Based on this goal,this paper has done some related research.The main work is as follows:(1)The gait detection link was main studied,including video object segmentation method based on background subtraction,gait detection method based on local binary mode,and edge detection based on differential method.Then according to the experimental analysis,and comparing the calculation amount and the influence of noise,the background difference method is used to detect the moving target,and the first-order differential method is used for edge detection.(2)Focus on the feature extraction of gait.By analyzing the problems of Histogram of Oriented Gradients(HOG),a method based on Pyramid Histogram of Oriented Gradients(PHOG)and fusion gait energy map was proposed.Specifically,after the gait detection is completed,a gait energy map is calculated,and the edge information of the gait energy map is extracted by a first-order differential method.Finally the PHOG feature vector of the gait energy map is extracted for gait identification.(3)The classification and recognition of gait was studied.By comparing the traditional classifiers such as SVM and Extreme Learning Machine(ELM),ELM has several advantages,such as fast learning speed,high classification accuracy and good generalization performance.Thus,ELM is utilized to recognize gait.(4)The PHOG-based gait recognition system algorithm is simulated by Matlab,and the simulation results are analyzed.In addition,through the above research,we found that:(1)By performing PHOG on the edge of Gait Energy Image(GEI),the extracted features have strong noise immunity and certain anti rotation ability,and are suitable for gait recognition.(2)The extracted hierarchical feature by PHOG not only represents the global shape and local details of gait behavior,but also describes the spatial information of gait behavior.