Towards collision alarming based on visual motion
Mehmet Kılıçarslan, Jiang Yu Zheng · 2012
This work models various collision situations that may happen to a driving vehicle on road in probability, and map such events to the visual field of the camera. The identification of dangerous events thus can be carried out based on the location-specific motion information modeled in the likelihood probability distributions. Depending on the motion flows detected in the camera, our algorithm will identify the potential dangers and compute the time to collision for alarming. With the location dependent motion based on the knowledge of road environment and behaviors of other vehicles, this approach will detect the motion of potential dangers directly for accident avoidance. The mechanism to link visual motion to the dangerous events avoids the complex shape recognition of vehicles so that the system can response without delay.