Gait Recognition Research Based On Feature Fusion Convolutional Neural Network

With the rapid development of the information society,people's safety in the public environment has been paid more and more attention.Among the many biometric recognition technologies,gait recognition is attracting more and more scholars' attention.Gait recognition uses people's walking posture to achieve the recognition of people.Compared with other biometric technologies,gait recognition can be used over longer distances.It has the advantage of long distance,non-contact,non-perception,and nonaggressiveness,and is difficult to camouflage.Therefore,the gait recognition has great application potential in the field of intelligent monitoring.In this article,we have conducted a series of studies around gait recognition based on convolutional neural networks.Aiming at the problem of insufficient dynamic information of the gait template GEI(Gait Energy Image),which is commonly used in gait recognition,we have proposed a Cross-View gait recognition method based on spatio-temporal feature fusion with a two-stream convolutional neural network.The proposed method introduces temporal information by constructing a gait temporal image(GTI)at the feature layer,combines it with the original silhouette,and use the proposed STGNet convolutional neural network to extracts the spatio-temporal features of the gait.With the help of the deep neural network,the proposed method can well extract the spatial and motion features of the gait across different angle.The proposed method achieves a high recognition rate on a large data set OU-ISIR Large Population,the proposed experiments verifies the feasibility of the proposed method,and show that the feature fusion algorithm can fuse the spatial and temporal characteristics of the gait effectively,which improves the recognition rate of the gait.It is more effective than the general GEI + CNN method.It has a higher recognition rate than the mainstream deep learning based gait recognition method.And shows certain robustness under perspective transformation.