| In recent years, research on passenger flow monitoring system has been being a very active field of computer vision and intelligent transportation. The monitoring system has been applied more and more widely. In public security through monitoring passenger flow can establish warning mechanism so that reducing accident occurred of swarm and jostlement. In commercial field commercial investor can timely adjust management strategy according to the change of passenger volume so that improving shopping environment and the rationality of the human resource allocation. In the field of transportation traffic route planners can formulate reasonable traffic lines through passenger flow data so that improving efficiency of transportation.Current passenger flow technology is often affected by many factors such as illumination changes and complex background scene. Especially under the crowded environment, mutual occlusion of moving targets will bring the great difficulty and challenge. In this case, the traditional passenger flow monitoring system is difficult to satisfy the practical requirements of accuracy and real-time. Based on this, this article designs and implements a system which can be applied to crowded environment.On the basis of researching some pedestrian detection algorithms, this article adopts the method of depth image and color image combination to detect moving pedestrians. It firstly uses the algorithms of combining HOG and SVM to identify targets. Then it adopts MeanShift combined with Kalman filter algorithm to tracking object so that obtaining tracking trajectory. After that, it counts pedestrians and computes velocity and density of passenger flow according to tracking trajectory. In addition to the function of passenger flow data computing, the system also realizes the functions of data conservation and remote monitoring. Finally, by the field test in a subway station, the stability, and accuracy of this monitoring system has been verified. |
PDF Full Text Request