Detection of Road Image Borders Based on Texture Classification

Abstract

The concept of an algorithm developed for the segmentation of a road border from the content of an image produced by a forward looking TV camera mounted on a moving vehicle is presented in this paper. The extraction of a road boundary is an important step in the context of autonomous vehicle guidance, enabling further calculations of distance from the border, direction of a road, etc. The main idea behind this approach is that the texture of a road is different enough in comparison to the textures characterizing the surrounding environment, allowing the separation of the overall image into a few distinguishable regions. The segmentation algorithm combines the texture descriptors of a statistical nature and the ones based on a grey level co-occurrence matrix. The significance of this work is mainly in the practical verification of the proposed algorithm and in the testing of the real limits of its application.