Research On Cross-Camera Target Recognition And Tracking Method Based On Face And Pedestrian Features

As a reliable and mature traditional technology,intelligent video technology has already become one of the main technical means in public security work such as criminal case investigation,public security management prevention and control,and traffic management.In the construction process of "Xueliang Project",the pursuit of quantity and hardware,neglecting quality and application,and inefficient utilization of collected data and information.In recent years,with the continuous advancement of "smart public security" and "smart security",pedestrian re-identification and face recognition technologies have been widely used in public security video surveillance integrated platforms.Face recognition is mainly used for target recognition and identity verification,both of which have become indispensable core technologies in public security work.Aiming at the problems of single monitoring target feature extraction,low utilization rate of collected data,and low system search efficiency in existing video surveillance,this paper proposes a target recognition and tracking system that integrates face features and pedestrian features.The search part of the system adopts An improved breadth-first algorithm based on spatial information is proposed,and the effectiveness of the algorithm and the search efficiency of the system are analyzed and verified through experiments.The main work and innovations of this paper are as follows:First,the fusion application of facial features and pedestrian features is performed.Both face features and pedestrian features have their own limitations in target recognition and tracking,and they are integrated and applied through certain strategies and calling methods.Second,the front-end smart camera has been improved.By adding a GPS embedded chip in the front-end smart camera,the camera can collect the physical location information of the deployment location and send it back when it is deployed.Thirdly,the physical location information of the front-end camera is associated with the video data information collected by it,and a spatial information database is established,which is applied to the improved breadth-first search algorithm in the form of nodes.By comparing with the traditional traversal algorithm,it is checked whether the performance of the improved algorithm is improved.Finally,an improved target recognition and tracking system is designed and implemented,which realizes the fusion application of facial features and pedestrian features,and improves the breadth-first search algorithm as a systematic search algorithm by combining spatial information.The ablation experiment analyzes the fusion application effect of face features and pedestrian features,and tests the effectiveness of the improved breadth-first search algorithm.