Detecting Highway Road/Lane Boundaries at Night

Abstract

Detecting highway road/lane boundaries is an important task for autonomous vehicle navigation. In this research, we study the case at night in which the bright spots of reflector plates indicate the road/lane borders. Through the brightness and area filtering in the process of detecting the bright spots, the noise interference caused by other vehicles and other light can be eliminated. The filtered bright spots are then sorted based on their locations on ground and approximated by curves of the second order polynomial on the ground coordinates. Based on our scheme, experiments have shown correct and reliable detection of the highway road/lane boundaries under various road scenes and noise environments.