Research On Human Flow Statistic System Of Dense Population

The real-time statistics of dense crowd flow in public places is of great significance to social management and public safety.In this paper,a human traffic statistics system is designed and implemented on the background of camera fixed angle,which can detect,trace and count the dense crowd in real time.The commonly used methods of moving target detection are: optical flow,frame difference and background difference.This study carried out experiments on all three methods,analyzed the advantages and disadvantages of the three methods and the applicable environment.A self-adaptive background extraction and improved three-frame difference method are used as the pedestrian detection algorithm.In the aspect of pedestrian tracking,the commonly used tracking methods are studied,and the advantages and disadvantages of several methods are analyzed.In the case of large population density and relatively severe occlusion,a pedestrian tracking algorithm based on multi-feature matching is adopted in this paper.For head detection,a head detection algorithm based on centroid matching is proposed.After preprocessing the detected video images,the image centroid matching index of pedestrian head is established.In the case of the three conditions of the nearest centroid matching,the cross boundary detection and the motion direction conformance,the location of the head is realized and analyzed.Finally,in order to verify the real-time performance and effectiveness of the proposed algorithm,we have selected arrays of different scenes and different numbers of video sequences to test the system.The experimental results show that the algorithm has strong anti-interference ability,the system can run in real time for a long time,and can realize the count of human traffic in many directions and the rate of statistical accuracy.It can achieve more than 92%.Effective and accurate traffic statistics can be achieved at a reasonable cost to meet actual needs.