ROI based real time straight lane line detection using Canny Edge Detector and masked bitwise operator

Abstract

Research for autonomous cars has now been close to a decade and still it is not possible to employ these cars everywhere around the world, for one major reason being clear lane line detection. However, there is constant discovery to improve the method of lane detection, especially in real-time. For lane detection, various computer-vision techniques and deep learning models have been devised, but for practical use it is necessary to find an efficient solution in real-time. Our technique is based on the real-time efficient detection of straight lanes using a canny edge detector followed by finding a region of interest and Hough transformation. This method takes video as an input and gives outputs in the form of images with slopes and marked lines of lanes. For long highways with straight lanes, this algorithm can prove to be extremely efficient for detection, which can be easily employed in real-time using camera sensors that provide a video feed. Furthermore, there is no requirement for training the algorithm. Hence, this system works on most of the scenarios without any prior data training.