Lane Detection Based on Object Segmentation and Piecewise Fitting

Abstract

A lane detection algorithm for complex environment was proposed. It was concerned on selecting candidate lane region by object segmentation. Then redundancy edges were extracted by Sobel operator. Furthermore, candidate lane markers were obtained by threshold selection from the edges. Finally lane markers were detected by piecewise fitting. The proposed algorithm was simulated in MATLAB. Experiments showed that lane markers could be detected correctly. Piecewise linear transformation in preprocessing has enhanced performance of detection while the environment was dim. And limited region of interest helps to identification lane in an appropriate region, which have the effect of enhancement in the speed of operation. Feature-based method is usually affected by intensity of image. Several characteristics of roads need to be considered in further for detection more precisely. DOI : http://dx.doi.org/10.11591/telkomnika.v12i5.3240