Lane Detection and Estimation using Perspective Image

Abstract

Lateral localization of an autonomous vehicle within its lane is major information for its adequate control and navigation. Computer vision and robotics communities have used primarily images to Bird's Eye View for easier data manipulation than perspective image. Nevertheless, this technique usually assumes that the terrain is flat and needs calibration for its transformation matrix. In this paper an efficient method of detection and estimation of lane-based perspective image of a monocular camera is presented. Our algorithm is based on robust image processing, using Probabilistic Hough Transform, road marker estimation, and vehicle lateral localization in the lane. Our system provides satisfactory results, demonstrating its ability to detect lanes in several situations, including in variable light conditions, and even during the night. Our system also does not rely on metric data, but provides useful control information using the pixel's proportion. Therefore, the proposed methodology contributes a robust and user-friendly system that depends exclusively on a perspective image.