Title :
Towards collision alarming based on visual motion
Author :
Kilicarslan, M. ; Zheng, J.Y.
Author_Institution :
Dept. of Comput. Sci., Indiana Univ., Bloomington, IN, USA
Abstract :
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.
Keywords :
collision avoidance; image motion analysis; image sensors; image sequences; object detection; road accidents; statistical distributions; traffic engineering computing; accident avoidance; camera visual field; collision alarming; collision situations; dangerous event identification; likelihood probability distributions; location dependent motion; location-specific motion information; motion flow detection; road environment; visual motion; Bayesian methods; Cameras; Feature extraction; Gaussian distribution; Probability distribution; Roads; Vehicles;
Conference_Titel :
Intelligent Transportation Systems (ITSC), 2012 15th International IEEE Conference on
Conference_Location :
Anchorage, AK
Print_ISBN :
978-1-4673-3064-0
Electronic_ISBN :
2153-0009
DOI :
10.1109/ITSC.2012.6338835