Vision Based Lane detection for Active Security in Intelligent Vehicle

This paper presents a monochromic camera lane detection algorithm based on Snake for improving active security in intelligent vehicle. The algorithm allows the elimination of several common assumptions: flat road, constant pitch angle or absence of roll angle, which are the limitations of many former algorithms. Moreover, compared with the former methods, the snake based lane model is able to describe any arbitrary shape by a set of control points. We use an improved Hough transform method to wipe off disturbances caused by shadow or edge of other things. Then we get the accurate vanishing points using a method of mutual comparison to desert some bad ones. The method of update and tracking is also referenced at last. The results of experiments, which have been done in urban road, show that the algorithm is adapted to many road conditions. Even in a complex driving environment, it also has a good performance.