To improve the real time performance and robustness, lane coordination detection is executed, which detects the lane parameter by adopting the global extraction based on Hough transformation and the local windows extraction with the help of the lane parameter confidence interval and the Kalman filter. Image preprocessing is conducted including the improved fast median filter, the lane feature filter, the gradient angular histogram selection, and the lane connectivity analysis to cut the computational expense and enhance the lane edge feature after the lane region of interest is defined. Real vehicle experiments show that the proposed method can run at an average speed of 25 frames per second on a Pentium (R) D 2.00 GHz CPU, and the accuracy can reach 97.2%. Meanwhile, it can robustly detect the lane even if there are some interference factors in the road such as shadow, vehicle and landmark etc., as well as fuzzy lane boundaries and relatively weak contrast.