The method of traffic flow detection is a key point of intelligent transportation systems (ITS). By the detection and recognition of video, traffic flow data, such as vehicle flow, road situation and so on, can be got so as to avoid traffic jam before it happens.This paper discusses the methods of traffic flow detection based on embedded system and image processing technology. The paper is consist of the following parts:Firstly, the whole architecture of system is given. On the way of modularization design method, the hardware platform is done. A new chip DMCU is used for the high performance and low cost of the system.Secondly, the algorithm of video traffic flow detection is researched. After the analysis of each algorithm, "Sub-background" method is adopt for the sake of embedded system. After simulation of the algorithm on PC, an improving method is turned out, named 'virtual check line and dynamic background'.Thirdly, the algorithm is ported to DMCU platform, then optimization of programming language and hardware is done to solve the shortness of embedded system.During the system designing period, a RTOS (KB OS) is used to manage the multi-tasks. It is helpful to decrease difficulty of the system.Last but not least, the analysis of system performance is give. It points out a good way to keep on.The whole system has been implemented successfully, and has been verified in the lab.Qian Liang(Control Theory and Control Engineering)Supervised by Tang Minghao...
|