Real-Time Image Processing System for Accident Avoidance

Abstract

Most of the vehicle calamity occurs during dark hours. This paper describes an analytical system for Vehicle Detecting, Tracking, Distance calculating and alerting the driver at night time. The vehicle lights are the salient visual feature to detect vehicle for the night conditions. Only the Red color channel and Symmetry of rear-lamp pairs are used to detect the rear lamps of the vehicles. RGB and HSV color spaces are used for threshold value calculation. Symmetry checking is performed by cross-correlation. Kalman filter is an approach for tracking, its advantage for the vision tracking is to tolerate small occlusions. Tracking based detection has been introduced and it is used when the symmetry check fails, this improves the robustness of the system in different lighting conditions and road environments. Finally the distance between the host vehicle and the target vehicle is calculated for the operating area. The driver has been alerted if the vehicle is in collision condition.