Research Of Real-time Detection Of Traffic Flow System Based On Video

These years, intelligent transportation system has recently received much attention due to the increase of traffic jams and the seriousness of accidents. The technology of video traffic information detection discussed in this paper plays an important part in the ITS. Through traffic image analysis and processing it can obtain many traffic flow parameters, they are much useful traffic information that is key part of the automatization traffic management.This thesis presents a real-time traffic flow detecting system based on the video and digital image processing technique. It first provides the background information of the video detection of traffic flow system, as well as the current research and application of this technology in our country and abroad, showing the significance and the prospect of this system. Then, it compares and analyzes the problems that exist in image processing, such as image filtering and noise reduction, enhancement. Then it establishes the basic method of calculation that will be adopted in the design and research of the system. After preprocessing, based on digital image differential detecting technology, we present a real-time method of vehicle detection. By dividing the detected object, extracting the edge information of the moving object. This paper uses a new detection method to extract lines which have vertical edge features. It solves the disadvantages of the traditional Hough transform, such as multi-line width and lacking of the information of the line dot. Based on location features of preceding vehicles in images, the line which is not the edge of vehicle can be removed from the three lines extracted by Hough transform . Then the vehicle's real edges can be determined by the remaining two lines. Finally, the vehicle's top edges can be recognized and located on base of vertical lines. Considering the static feature of moving object, we can sort and differentiate from people, bicycle, motorcycle and motorcar by figure information such as posture ratio (width- height ratio), Then filtering the moving object except the motorcar.This paper introduces the main composing configuration on the board of the programmable digital media processor DM642 and the soft development environment CCS(Code Composer Studio). Finally, it presents the design of video detection of traffic flow system. Then the vehicle detection algorithm is discussed in detail. Vehicles can be detected with the virtual line. Through the method of virtual line detection many traffic information can be achieved. The information consist of vehicle discrimination, vehicle number detection and velocity detection. And then algorithm flow is described. In order to improving the real time performance of the system, the thesis introduce some methods about system optimization.At last, the thesis gives a summary of my work and points out some problems and ideas for further research.