LANE DETECTION USING SUPPORT VECTOR MACHINES

摘要

Understanding lane is an essential step to provide morernrealistic information for video-based navigation systems.rnIn this paper, we present a novel idea to understand lanernfrom a live video captured in a moving vehicle. Morernspecifically, 1) lane markings are extracted first. Then, 2)rncolor information of lane markings are fed into supportrnvector machines to decide if it is yellow lane or not. Byrncombining information from database, it is possible torndecide if we are in the leftmost, middle, or the rightmostrnlane, which allows us to provide more realistic navigationrninformation to drivers. Exhaustive simulation results arernprovided to show the robustness of the proposed idea.