Abstract

Lane detection is one of the most important components in autonomous vehicle technology, driverassistance systems and lane departure warning systems. In this paper, we propose a new approach for efficient lane detection and tracking regardless of the varying illumination and road conditions. Our method is based on the use of a robust feature extraction and tracking method, however, its main novelties are the use of an adaptive and natural image enhancement technique based on global tone mapping using a modified gamma correction in order to render our algorithm robust in the nighttime road scenes as well as in the daytime, and the use of a 2D B-spline curve model to effectively and accurately detect and fit lane marks. Particle filter is then used for a robust tracking of the lane model. It can effectively complete lane detection and tracking in complex environments. What's more, we proposed a new scheme, which performs the detecting algorithm on a specified Region Of Interest (ROI) thus improving computational efficiency by reducing by a big amount the computing time. Our algorithm also proves to be robust in critical shadow conditions and cope with situations where lane markings are partly missing or obscured. The proposed algorithm has been tested on road images obtained by a driving test and its efficiency has also been verified.

Paper Details