AVM 영상에서 강인한 주행차선 마킹 검출 방법

초록

In this paper, we present a robust ego-lane detection method for various road environments. Conventional lane detection methods based on image processin are vulnerable to weather, traffic conditions and road markings whose width is similar to those of lane markings. In our proposed method, information obtained from AVM"s left and right images are utilized simultaneously with additionally processed filter response from specifically designed filter for lane detection. We confirmed experimentally that the proposed method efficiently detects lanes while rejecting other road markings.