A novel algorithm of lane detection addressing varied scenarios of curved and dashed lanemarks

Abstract

Lane detection is a critical step in advanced driver assistance systems (ADAS). The detected lane information is used by later modules of warning and controlling the differential brake and steering angle. Here we propose an efficient algorithm for detecting accurate lane inbounds under varying illumination and road conditions like curvy, straight and dashed lane markings, deterministically. The current method uses edges extracted from frames and lane mask information extracted from spatial lane stripes through adaptive figure-ground segregation to determine the lane bounds. The proposed method has been extensively validated for varying real time scenarios against the state-of-art methods of lane detection.