Line labelling and region segmentation in stereo image pairs

Abstract

This paper presents an algorithm for extracting regions and subsequently labelling line segments in images, the algorithm is implemented for a stereo system that consists of a verging pair of cameras mounted on a mobile robot. The focus of the paper is on the description of an algorithm that uses the geometrical layout of line segments in an image to attach a label to each line segment: boundary, join or isolated. This line labelling is then used in conjunction with the planar regions within each image of a stereo pair to facilitate high level stereo matching.