Combination of the symmetrical local threshold and the sobel edge detector for lane feature extraction

Abstract

In this paper, two of lane feature extractors are combined for better lane feature extraction. Namely; the Symmetrical local threshold (SLT) and the Sobel edge detector. The SLT is found to be one of the most robust lane marking feature extractor. However, it relies on the Dark-Light-Dark feature of the painted lane markings. However, in many cases there is no painted lane marking on one or two sides of the road border. Although, edge detector is not assumed to be as good as the SLT for the lane marking feature extraction, it has more advantage on detecting the road borders. Thus, even though many lane marking feature extractors are proposed in the literature, edge detectors are still popular method for lane detection. In this paper, a new approach to combine two of the feature extractors is proposed. With the proposed approach, while detecting painted lane markings accurately, it is possible to detect road borders and poorly painted road markings as well.