Comparison of Edge Detection Technique for Lane Analysis by Improved Hough Transform

Lane detection system for car driver assisted becomes an important study to be implemented for safety purposes. It used to lessen possibility of traffic accidents, to monitor the position of a car effectively and to contribute for further development of autonomous navigation technology. In this paper, we proposed an improved Hough transform technique to detect road lane where a comparison has been made on edge detection technique of Canny, Sobel and Roberts. The improved Hough Transform used to extract the features of structured roads. The close field-of-view scope adopts a straight line model to accelerate the speed of data calculation and to find the fitting line. Prior-knowledge is used in lane finding process to efficiently decrease Hough space efficiently, thus enhancing its robustness by improving the processing speed. The algorithm gave good result in detecting straight and smooth curvature lane on highway even when the lane was affected by shadow. The data of road lane has been taken in a video format. Experiment has been done by making a comparison of edge detection technique and we find that the best method that produces high accuracy of detection is by using canny edge detector.