The Study Of Traffic Flow Video Corner Detection Technology

Traffic plays a very important role for economic development and the improvement of living standards. Intelligent Transportation Systems(ITS) represents the development direction of modern transportation, which is a real-time, high efficiency, omni-directional comprehensive transportation management system. To construct a complete ITS system, first, a traffic flow parameters detection system should be established, which can get traffic characteristic parameters accurately and efficiently, so this paper did research on traffic flow parameters detection methods.This paper proposed a detection method through analyzing the existing traffic flow parameters detection methods, designed and fullfilled the the system frame model, functions and workflow. The paper focused on corner detection technology, analyzed the existing various corner detection algorithms, and finally proposed an Improved SUSAN Corner Detection algorithm(ISCD). On the basis of SUSAN algorithm this algorithm used adaptive threshold extraction and quick corner detection, succeed in avoiding manual setting threshold and improving the running speed. Based on target detection, calculate the actual length of the vehicles for using ISCD algorithm and camera calibration technology, further calculate space occupancy of the road, which can be used as the basis of measuring road running status.The paper also studied the detection methods of other traffic flow parameters such as traffic volume, vehicle real-time speed and vehicle average speed. Traffic flow parameters detection is based on target detection and tracking, and the paper selected appropriate algorithm to achieve, including the background extraction and updating algorithm, target detection algorithm, shadow elimination algorithm and target tracking algorithm. After testing, the traffic flow parameters the system detected all have a certain degree of accuracy, and can satisfy the basic request of system design.Based on these studies, the paper developed the system of Traffic Flow Parameters Detection, including detection program running on the video detection server, detection client and drawing software. The system runs normally, can reach the expected study goals.