Sensing of Automotive Environments Using Stereo Vision

This paper presents a vision system to be used on moving vehicles to increment road safety. Pairs of images acquired by two stereo cameras installed on the vehicle are processed in order to detect both generic obstacles on the path of the vehicle and the road position. Both stereo images are used for the detection of the free-space in front of the vehicle, while only the left one is used for the identification of the road, which is performed through the detection of lane markings. The two functionalities share the same underlying approach, thus easing the fusion of the results of the two processings. The system has been tested both in laboratory and on board of MOB-LAB experimental vehicle and has demonstrated to be robust with respect to shadows, different road textures and vehicle movements. INTRODUCTION This paper presents a stereo vision system for the detection of obstacles and the identification of the lane in a road environment that processes grey-level images acquired by a pa...