In recent years we notice a sudden boom in vehicle population, this sudden swell is also the cause of the increase in the number of road accidents. Unintended lane departure is the major cause of these accidents which endanger the lives of people. A proper lane departure warning system can prevent these accidents and significantly improve the safety of the driver and others. In this paper, we use a computer vision-based image processing approach to tackle the problem. The proposed system is robust and capable of achieving real-time operation. The road view is captured by using a camera attached to the dashboard of the vehicle. The input video is processed frame by frame to determine the departure of vehicle from the lane and generate a warning signal. The pre-processing phase involves detection of Region of Interest ROI then applying filters to remove the aberration caused by shadows. After this gray scaling is performed on the ROI and it is finally converted to a binary image using Otsu's threshold method. Edges are extracted from the obtained binary image. Hough transform is then applied to the edge image and lane markers are determined. The advanced Hough transform algorithm is implemented to reduce the computation time and make system effective for real time application. The computation code is developed in MATLAB software and tested for different sets of images under different lightning conditions in the presence and absence of shadows. The accuracy of lane detection is found to be 97% for still images. To simulate the real time driving environment, frame by frame analysis of the video captured by a camera mounted on the dashboard of a running car is performed. Time taken to analyze a video frame is ∼ 82 ms. Out of the total 1865 video frames analyzed, the number of false detection false positive + false negative is less than 3%. Performance of the system is found to be efficient and accurate for use in real time applications. Computation algorithm with test results is presented in this paper. © 2018 IEEE.