Fast Hough Transform for Road-Line Detection

This paper proposes a novel Hough transform method that is suitable for road-line detection. The Hough transform is the most efficient and intuitive way to detect straight road lines. Previous methods for detecting straight road lines have used the general Hough transform. If the range of the vehicle’s motion and the camera’s position are restricted, the line candidate region in the road image can be limited to a narrow range. We propose a novel Hough transform method considering these characteristics. Our proposed method efficiently generates the Hough space, reducing the memory usage and generation time. Because of the error-filtering effect of the proposed Hough transform, we can reduce the detection error and detection time. Compared with previous Hough transform methods, the accuracy of the proposed method is the same or higher. The proposed Hough transform method, which has the advantage of lower memory usage and short computing time, is quite suitable for embedded systems.