Multi-target Tracking System For Multi-camera

With the Wisdom City,Safe City and other large-scale intelligent information management system starting the layout in major cities at home and abroad,people for the security industry requirements continue to improve,intelligent monitoring system development is particularly important.The multi-target tracking system of the camera is an important research content in the field of intelligent monitoring system.It aims to realize the continuous tracking of the moving target under different cameras,which has important research significance and broad market application prospect.The cross-camera multi-target tracking system is an important research content in the field of intelligent monitoring system.It aims to realize the continuous tracking of the moving target under different cameras,which has important research significance and broad market application prospect.Cross-camera multi-target tracking system involves a number of aspects of research,such as target detection,the multi-target tracking under single-camera,and target correlation.Due to the interference of real factors such as background occlusion,illumination change and camera angle difference,the multi-target tracking algorithm is faced with great challenge.Therefore,the study of the system has significant scientific research value.In this paper,we first introduce the basic theory and algorithm of dynamic target detection and multi-target tracking,and then propose an algorithm to solve multi-target tracking only using the dynamic information of target motion.Finally,we introduce the relevant modules in system design and the related targets detection,tracking and association algorithms.The main research work and innovation of this paper are as follows:(1)A multi-target tracking algorithm based on moving motion of target and multi-layer hypergraph is proposed.First,without using the motion model,the motion dynamic information of all the targets in the whole video is obtained by using the target motion dynamics,and the similarity comparison function between.the track fragments is constructed to reduce the mismatch between different objects with similar appearance;Then,By using the method of hypergraph association,the global search clustering of each trajectory fragment makes the tracking problem become a hyper edge set problem of a dynamic search hypergraph.The hypergraph association of the optimization process fuels our tracker to long-term occlusion with more robustness.Experiments show that in the challenging common video sequence,the method presented in this papershows its good superiority:can handle complex target motions,non-stationary cameras and long occlusions,on scenarios where appearance cues are not available.(2)To design cross-camera multi-target tracking system.The system mainly includes two parts:the tracking module and the correlation module.The system mainly includes two parts:the tracking module and the correlation module.The tracking module is used to detect and track the moving target under the single camera.Where the target detection movement is based on the improved HOG and Sobel-LBP fusion fast pedestrian detection algorithm,and target tracking uses a multi-target tracking algorithm based on adjacent frame feature matching.Experimental results show that the combination of the two can be more effective to achieve a single camera under the target tracking.In the correlation module,firstly,the time and space model between multiple cameras is established,and the BTF function is used to suppress the illumination changes of different cameras.Under the constraints of the space-time relationship model,the optimal match of the target under different cameras is realized by the Hungarian algorithm In order to achieve cross-camera multi-target tracking;Finally,the interface of the two modules are shown.