Design And Implementation Of Binocular Bus Passenger Flow Statistics System Based On OpenCV

The development of urban public transport industry has a great influence on the comprehensive development speed of a city.The development of public transport is also one of the important indicators to evaluate the urban planning level.More and more cities are beginning to pay attention to the development of public transport.Bus dispatching systems have started to popularize in major cities,yet few cities can monitor the number of passengers on the bus in real time and the number of passengers on board.The traffic statistics system can record the number of boarding and alighting passengers in each station at actual time and transmit the data to the bus company and the transportation management center to provide a scientific and effective data reference for the bus scheduling,being conducive to the improvement of bus routes,the increase and decrease of numbers of buses,so as to improve the overall utilization of bus lines and rationality.The traditional passenger flow statistics scheme no longer meet the increasing traffic statistics demand due to the increase of labor costs and the low statistical accuracy and stability.This thesis presents a binocular passenger flow statistics system that can automatically calculate the number of bus passengers in real time and make up for the shortcomings of the traditional passenger flow statistics.The main research contents of binocular vision passenger flow statistical system proposed in this thesis include the following points: using the binocular imaging principle to generate the depth image of the bus passengers;post processing the depth map so that the image can facilitate the extraction of the passenger target;proposing a method of locating the head of a passenger based on MSER algorithm combined with feature selection;proposing a motion estimation method based on Kalman filter to estimate the motion of a passenger.The centroid algorithm and regional template matching are used to obtain the motion trajectory of the passenger;Designing a counting statistical line which can be used to calculate the passenger flow by analyzing the relationship between the track of passengers getting on or off and the statistical lines.All image processing programs in this thesis are based on OpenCV.Experiments reveal that this method can meet the real-time requirements of the premise,the overall accuracy rate of 95%,in line with the needs of the number of urban public transport statistics.The hardware platform used in this thesis is the Hi3531 chip,which realizes the migration of the algorithm under this platform.The findings of the test of this system equipped with the buses indicate that the system has high accuracy,well robustness and can meet the needs of urban bus traffic statistics.This system can also adapt to the future development of intelligent bus needs.