Research And Application Of Traffic Video Analysis Techniques For Intelligent Transportation System

This paper makes an effort on a systematically study of the foremost image processing techniques with their development and applications utilized in traffic video analysis. To accomplish this goal, the paper firstly explained the Intelligent Transportation Systems (ITS) to a detailed level in its general development, function designations and detection requirements for video analysis part. Then it came to a fully introduction to the supposed modules and sequenced procedures for typical video analysis sub-system in ITS, and also some image processing approaches they indispensably rely on. On basis of them, this paper attempts to analyze each step of traffic video analysis in their functions and gathers most maturely developed theories and algorithms others have discussed for accomplishing these tasks. The discussion mainly includes those algorithms for motion detection and segmentation, shadow discrimination, vehicle detecting, matching, and tracking. It also includes some approaches for obtaining important traffic information like the traffic flow, with indirect methods such as remodeling the camera view.For each procedure of the video analysis system, this paper has made comparisons between many suggested algorithms on their advantages, disadvantages, limitations and robusticity. By the unique features of each analysis step, this paper improves, synthesizes and creates algorithms to satisfy the requirements of traffic detection, which are being more and more fastidious. At the last part, this paper suggested a few problems in experimenting the detecting system in roadside surveillance and gives some theories and algorithms the author has applied to solve the problems.