Abstract
Moving Object Tracking (MOT) is one of the frequent used tasks in computer vision systems and on the same time a challenging issue particularly in darkness. Vehicle tracking involves finding new position of vehicles in consecutive frames. This paper presents MOT algorithm that is developed for advanced driver safety applications like automatic high beam control, forward collision warning. Accordingly, the proposed approach targets vehicle tracking in the dark environment. Hence, a camera is mounted on the host vehicle to capture video frames of the traffic ahead. The scope involves tracking of both oncoming and preceding vehicles. The vehicles are tracked in consecutive frames using grayscale information and robust structure features. The features representation of the vehicle region is based on cumulative pixel intensity information. The implementation for feature extraction is optimized by using a dynamic programming approach to meet the constraints of a real time application. Simulation results thus obtained are promising in state of the art.