Tailgating enforcement based on back-tracking in intersection

Abstract

In this paper, we propose the new method to enforce the tailgating violations that occurs at the intersection. The tailgating enforcement system is composed of the wide-view camera for tracking a vehicle and the narrow-view camera for recognizing a license plate. The images from the wide-view camera are sequentially stored in the stack in order to monitor the ROI (Region of Interest) at the intersection area. The narrow-view camera recognized the license plate. And the image coordinate of the license plate from the narrow-view camera was converted into the image coordinate from the synchronized wide-view camera. The feature points of the vehicle were extracted the image from wide-view camera using the converted image coordinate of the license plate from the narrow-camera. The images from the wide-view camera are searched in reverse order, and the vehicle trajectories are tracked. As a result, the success rate of the back-tracking is 98% at day and 93% at night. And the success rate of the sequential tracking is 90% at day and 76% at night. The back-tracking is the better results than the sequential tracking. Therefore the back-tracking is the appropriate to enforce the tailgating violation at the intersection.