A Method for Traffic Collision Detection
Dong Yin, Fan Zhang, Kun Wang · Applied Mechanics and Materials · 2012
This paper presents a detection method for traffic accident in real-time video images. Firstly, according to improved average background model, frame difference method and edge detection technology are used to detect vehicles. Secondly, vehicle tracking is accomplished by matching the distance, area and histogram of the same vehicle in next frame. Finally, using the concept of collision area and key point as pre-qualification, the situation of vehicle collision will be accurately detected by prior knowledge and histogram information. The experiments results show that our method is effective.