In this paper, we propose a real time method to detect vehicles on road by using a vehicle mounted monocular camera. Based on lane detection design, low time cost and high accuracy vehicle detecting and tracking algorithm is achieved. Robust and fast lane detection has been achieved using Hough transform in combination with a line merging method. The lane data are used for effective vehicle hypothesis generation. A subsequent validation, based on the area ratio of hypothesis vehicle, is used to eliminate false positives. Further, a data fusion mechanism is proposed to incorporate temporal information for stably updating vehicle detection results over time. Experimental results show that, without specific hardware and software optimizations, our method is able to detect vehicles on road with low false alarm rate at real time speeds of 30 frames per second.