Intelligent Traortation Systems (ITS) assist the driver to increase vehicle safety and to control (navigate) the vehicle independently. Lane Departure Warning Systems (LDWS) are a group of these systems that are designed to provide an audio alarm in case the car moves out of its line because of driver drowsiness. In this paper, a novel vision-based lane detection technique is proposed to detect the lanes in video frames with different lighting conditions, various types of lanes, and complicated road surfaces. The video frames are captured using an Apple iPhone 4s and Nokia N8 cameras(30 fps) that is mounted behind the windshield of a passenger car. Drowsiness mostly happens in highways and roads which are straight and not crowded. So in this paper we assume that this system is used when the car is moving in such a road or highway and it is also assumed that there are lane marks on the road. Since the camera position is fixed in an installation, left and right lane marks can be expected to be present in fixed regions in the captured images. So the processing is restricted to these regions to speed up the procedure. Canny edge detector is used to find the edges of the input frame and connected edge components are extracted. Left and right lane marks are selected from these components according to the position, orientation and pixel intensity around these components. Existence or nonexistence of left and right marks and their position in present and some previous frames are used to detect lane departure. Effect of some previous frames and car velocity are investigated in recognizing lane departure. Our goal was to develop an algorithm that could be utilized on a smart phone or tablet, so the algorithm steps are tried to be as simple as possible. The algorithm was tested on 6 videos with 7828 frames in different roads. There was 28 lane departure in these videos. All lane departures were detected and there was one false alarm in this test. The algorithm showed to be robust to camera installation and car speed. This algorithm was implemented in real-time (4 fps) on a tablet. Keywords: Intelligent Traortation Systems, Lane detection,weak processor,real-time.