关键词: 车道线跟踪, 车道线检测, 方向可调滤波器, 智能车

Abstract: Considering the directional characteristics of lanes, an image preprocessing method using steerable filters is proposed. Based on the analysis of lane orientation characteristics in images, images are divided into subareas, in each subarea the steerable filters which accord with the lane direction in that subarea are adopted. During the initial lane detection phase, edge distribution function (EDF) is used to determine the lane direction; during the lane tracking phase, the last-round the detection results are inputted as the orientation angle of steerable filters. Comparison shows that the method can intensify lanes information effectively and eliminate image noise. Experiments under various conditions show that the lane detection method using steerable filters can detect lanes stably, estimate lane parameters accurately, and the algorithm is fast enough.

Key words: Intelligent vehicle, Lane detection, Lane tracking, Steerable filters