Efficient Vanishing Point Detection for Advanced Driver Assistance System

The lane and driving direction keeping technologies are an essential technology for safety driving support in an autonomous vehicle. In particular, the vanishing-point detection technology that analyzes causal relationships with the surrounding objects using image analysis techniques and obtains additional information on lanes, road boundary lines, and other vehicles has been gaining attention in the autonomous vehicle research. The vanishing point is the point at which the parallel extension lines of objects converge in a three-dimensional space. Several useful functions can be applied to this point, including the prevention of deviation from lanes, distance estimation, and roadway detection. In this paper, we propose a method that detects the vanishing point from a road image obtained for the application of the black box, which is an image-storage device for cars. In the proposed method, an image’s edge and regions that require visual attention in directional characteristic information are detected, and the vanishing point is detected in real time via Hough transform and image segmentation. The proposed method was experiments in a variety of environments, and shows a relatively rapid detection results.