Structure The Path Of Multi-way Street Line Detection Algorithm

Vision-based driver assistance system has become a most important topic in the fieldof automated guidance vehicle recently. The lane detection is the indispensable part inautomated vehicle system, and the characteristic of real-time and stability in identifyinglane position is the foundation of decision-making and environment-recognition.The objective of this paper is to research the lane detection technology in daytimeand nighttime arid design the image process, including image grayed, edge detection,image binary, vanish point detection and lane detection.To solve the problem that image quality is influenced easily by light. This paperdevelops a heuristic search algorithm based on linear characteristic, the algorithm alsobinarizes grads image, which makes it excellent to binarize the road image in daytimeand nighttime.The paper utilizes the fact that parallel lines in space intersect at a point, obviouslines are detected by means of Hough transform, successively the vanish point could bedetected. On the basis of vanish point, this paper develops a novel model and analgorithm of lines detection constrained by the vanish point so that all the lanes in theimage can be detected by using a rapid clustering algorithm in Hough space.The experiment results of 365 road images in daytime and 570 ones in nighttimeprove that, the lane detection algorithm developed in this paper is stable and accurate inlane detection, which could be applied in real project.