Design And Implementation Of Passenger Flow Statistics Based On Detection And Tracking

With the progress of society and living standards, people’s business activities gradually increased. For some public places such as shopping malls, museums, airports, bus companies, subway stations, and tourist attractions, accurate and real-time traffic information are needed to provide the basis for their management and decision-making. Meanwhile, the rapid development of computer vision and pattern recognition, the traditional passenger flow statistics method cannot keep up with the times. Under this background,this paper designs a passenger flow statistics system based on detection and tracking.In this paper, we focus on multiple target detection and tracking algorithm. With the investigation of the target detection and tracking algorithms, this paper use the C++ language to implement the detection and tracking algorithm, and embeds the algorithm to our passenger flow statistics system. Based on the analysis of actual situation in test, we get the optimized passenger flow statistics system, which has better real-time and robustness at the same time, it can accurately detect and track object in complex situations. The proposed detection and tracking module are respectively embedded into the monitoring system to achieve the multiple moving targets real-time monitoring under the complex background and realizes a kind of sticky type passenger flow counting system based on detection and tracking finally.In this paper, the main research contents are as follows:(1) This paper give an introduction of a variety of target detection and tracking algorithm in detecting and tracking process. Then, it compared optical flow methods, frame difference methods in motion detection, and background subtraction algorithms for a simple description. It also analyzed the SVM and Adaboost algorithm, and compared the commonly used descriptors such as Haar, HOG and SIFT. We tried to find a detection algorithm that is suitable for the actual situation.(2) A pedestrian tracking method based on Supervised Descent Method(SDM) is adopted and embedded in the proposed system. The re-judgment mechanism is also introduced to solve the problem of drift in the target tracking, which can improve the tracking stability to a certain extent.(3) In this paper, we find a real-time accurate multiple target tracking algorithm and design an algorithm framework. Then, it realized under the environment of Visual Studio 2008 with C++ language. According to the requirements of practical application, we test the algorithm in the actual environment and optimize the algorithm based on the test results.(4) To realize a passenger flow statistical system, we embed the detection module and tracking module into the proposed system. It can guarantee the rates of error detection and false detection in the process of detection under complex conditions. Then, this system can realize a long time continuous tracking for the detection of multiple moving target, which will make the whole system with broad application market and economic value.