Recent years,lane detection has become of high interest in the area of intelligent vehicles and it provides the fundamental information which can be applied to the further development of Driving Assistance System.In this paper,we propose a lane detection system accurate to pixel level.It is based on iterative searching and Random Sample Consensus (RANSAC) curve fitting.First,we adopt fixed weight of RGB values to produce an intensity image,and gradient information is used as lane feature to get the edges of the image.Then,probabilistic Hough Transform is employed to estimate the probably areas of lane markings,and point by point searching with iteration is performed on these probably areas to retrieve an accurate set of lane pixels.Finally,we apply a fast RANSAC curve fitting algorithm with a novel score system to shape the lane markings.Experimental results show the effectiveness of our approach which detects lanes at a high rate in approximately 60 ms under different lighting conditions.