Incident Detection in Heavy Traffics in Tunnels by the Interlayer Feedback Algorithm

In this paper, we developed video surveillance system to detect incidents from heavy traffics in tunnels. Incidents in tunnels may cause additional accidents or fires that may result in fatal disaster. Therefore, it is important to detect incidents as soon as possible to manage the traffics inside the tunnels by the officers. Generally, video images in the tunnels suffer from heavy occlusion due to the low position of camera settings. In particular, the problem of heavy occlusions would be more serious in the urban tunnels due to their heavy traffics. We developed a tracking algorithm to segment vehicles and estimate the precise vehicle trajectories against the heavy occlusions. Utilizing this tracking algorithm, dedicated algorithm to detect incidents from the traffic images was developed. For the experiments, video streams of three cameras for 6 months were investigated, and 32 incidents were examined to evaluate the developed algorithm.