Research And Implementation Of Person Re-identification Based On Local Feature Fusio

Today,with the development of computer vision field,person re-identification has penetrated into all aspects of life.Person re-identification refers to the technology of finding the same person from images or videos taken by different angles and different cameras.It is mainly used to retrieve specific targets from a vast number of persons.In the research in this field,there is a problem that local feature information is difficult to fuse,which is also one of the main challenges in this field.In order to solve the above problems,this paper studies person re-identification based on local feature fusion,proposes a fusion method of spatial local features and temporal local features based on graph convolution,and implements a person re-identification system based on these.The research contents of this paper are as follows:In order to solve the problem that the relationship between parts cannot be obtained in the traditional part-based person re-identification method,a spatial local feature fusion method based on graph convolution is proposed.The spatial correlation between parts is found through human key point recognition,a relationship matrix is constructed,and graph convolution is performed according to the matrix to fuse the features of each part to obtain better global image features.In addition,in order to solve the problem that it is difficult to obtain the motion features of the temporal dimension in person videos,a temporal local feature fusion method of video based on graph convolution is proposed.This method uses a single frame as a node of the graph,finds the temporal correlation between frames through a self-learning network,builds a relationship matrix,and fuses the feature information of each frame through graph convolution to form an overall feature of a person.Finally,based on the person re-identification algorithm proposed in this paper,a person re-identification system is designed and implemented,which can run the above-mentioned person re-identification algorithm to find the most matching person in the database for the input person images and videos.