Lane detection is one of fundamental but critical problems for lane following system of Intelligent Vehicles. However, a robust and cost effective approach is still a deserve exploit issue. A novel and effective approach using a five steps scheme is presents. First, Canny detector is used to obtain edge map from the road image acquired from monocular camera mount on vehicle; Second, a matching process is conducted to normalize the potential candidates of road line or boundary; Third, a searching procession is utilized for reinforce potential road lines while degraded those impossible ones; Forth, a linking condition is investigated to further enhance the confidence of the potential lane lines; Finally, a K-means cluster based algorithm is employed to localize the lane lines, in this step, false edges will be rejection and only optimal lines will be accepted. Experimental results show that the proposed approach can achieve robust and effective localize the lane line in various bad road scenes, and has a better generalize capability with road types.