Intelligent Automatic Detection of Traffic Rule Violation From Traffic Surveillance Video Cameras

Checking traffic rule observation by vehicles play an important role in every day transportation handling either for intra- or inter- city travels. One of the most advance ways for traffic management and rules observation studies is employing live surveillance camera videos. In this paper, a new approach toward automatic detection of traffic rules violation based on image processing techniques is proposed. The proposed method applies innovative image processing techniques for live traffic surveillance target. Based on these techniques, the moving objects including cars and pedestrians are detected, tracked and observed. At first, some preprocessing steps employed for discrimination of foreground from background of surveillance video frames. For tracking purpose, a modified Munkres version of Hungarian algorithm is applied to Kalman filtering to provide tracking predictions for detected moving objects. In addition, the speed of tracked moving objects is detected based on optical flow related algorithms. The tracks of detected moving objects are analyzed and if any traffic rule violation takes place, they will be detected and reported automatically. The implementation results related to the proposed method demonstrates its high performance and applicability for real traffic rule violation detection

Keywords:

Gaussian Mixture Model , Kalman Filtering , Hungarian Algorithm , Intelligent Transportation System , Munkres’ Assignment Algorithm , Optical Flow