Road obstacle detection using disparity map and edge information

Abstract

In this paper, we propose a road obstacle detection method using a disparity map and edge information. Though we use a dense disparity map to improée obstacle detection performance, it is not easy to detect road obstacles accurately because of inaccurate disparity map. Furthermore, it is more difficult to detect long distance obstacles. Thus, we use another good segmentation cue, namely edge information. And we also propose a height limited column detection method to detect only certain road obstacles. Our method consists of three stages, namely feature extraction, obstacle detection using the height limited column detection and obstacle segmentation using a disparity map and edge information. To éerify our method, we conduct experiments using real traffic images captured by our stereo éision system installed in a real éehicle. © 2012 IADIS.