Title: Detection of irregular ground area by single camera on mobile vehicles

Authors: M. Tanaka; S. Onishi

Addresses: Department of Intelligence and Informatics, Konan University, 8-9-1 Okamoto, Higashinada-ku, Kobe 658-8501, Japan. ' Fuso Seiko Co. Ltd., Ibaraki, Osaka 567-0058, Japan

Abstract: For moving robots of outdoor use, it is essential to detect the movable ground area. In this paper, we propose a new method of making distinction between points on the ground surface (on-ground points) and not on the ground surface (off-ground points). These points are detected as feature points by SURF algorithm. In the model building phase, we build the projection transformation model of the ground seen from the camera angle. In the implementation phase, the camera detects feature points by SURF, and then makes correspondence of feature points between the consecutive frames. By inverse projection, the points can be classified into on-ground and off-ground points. This phase is executed online. We propose the principle of this theory and show an example to validate the theory.

Keywords: ground detection; mobile robots; single camera; computer vision; homography; SURF algorithm; irregular ground; projection transformation models; robot vision.

DOI: 10.1504/IJAMECHS.2012.048396

International Journal of Advanced Mechatronic Systems, 2012 Vol.4 No.2, pp.103 - 111

Published online: 30 Aug 2014 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article