In driving assistance system (DAS), road marking’s data can provide important assistance for driving safety. As the input image usually includes unnecessary information, lane detection system usually needs to remove most unnecessary data except for the lane markings. In this paper, a road marking extraction method is proposed to separate the painted lane lines using a special designed band-pass filter. Then we tried the acceleration method based on the integral image as well as the GPGPU. Road markings can be segmented automatically and adaptively in every frame. The proposed method is applied to various video images from black box, and is verified to be robust.