On The Identification of Snow Movements on Roads

One of the key components of an intelligent transportation systems is video-based automatic incident detection (AID). An AID system is able to detect incidents that require operator intervention. However, the accuracy of an AID system operating during the winter suffers from high false alarm rates due to the movement of snow on the roads. In this paper, a robust algorithm is proposed to detect moving snow in video streams and improve the rate of detection by having the AID system to reduce its sensitivity in the area that has snow movement. The proposed algorithm conducts glare processing, background generation & differencing, snow sample correlation and final snow map generation. The feasibility of the proposed algorithm has been evaluated using traffic videos captured from several cameras from the City of Calgary. This algorithm demonstrates accurate and real-time detection of moving snow