Intelligent Driver Assistance System Based on Panoramic Video Technique

Abstract

The intelligent driver assistance systems, which can warn a driver of possible collisions, would give the driver enough time to avoid the impact of collisions. Recently, video technique has obtained more attention in the intelligent driver assistance systems. This paper presents a panoramic video system using a pair of panoramic annular lens (PAL) cameras. PAL can give a panoramic view of the surround of the vehicle. Binocular vision analysis technique is used to detect the objects in front of the vehicle. Area-based correlation is then used to match corresponding features between images from two panoramic cameras. A disparity image obtained can be used to recognize the features and compute the distance from the vehicle. Each view of the vehicle’s sides is obtained from only one panoramic camera. In order to detect the objects, the consecutive frames from a panoramic camera are used to create a virtual binocular view. This produces a disparity image which also includes the features of objects and the distance. The result shows the basic feasibility of monitoring the surround of the vehicle based on the panoramic video technique. It is believed that panoramic video technique will play an important role in the future of intelligent driver assistance systems.