A Feature Extraction and Correspondence Algorithm for Laser Range Finder with Sensor Uncertainty

孙英杰,曹其新,李杰

DOI: https://doi.org/10.3969/j.issn.1005-9113.2004.04.003

2004-01-01

Abstract:This paper presents a feature extraction and correspondence algorithm which employs a novel feature transform. Unlike conventional approaches such as Hough Transform, we employ a robust but simple approach to extract the geometrical feature under real dynamic world conditions. Multi-threshold segmentation and the split-and-merge method are employed to interpret geometrical features such as edge, concave corners, convex corners, and segments in a unified framework. The features are represented by feature tree (F-Tree) so as to compactly represent the environments and some important properties of the F-Tree are discussed in this paper.To demonstrate the validity of the approach, we show the actual experiment results which are based on real Laser Range Finder data and real time analysis. The comparative study with Hough Transform shows the advantages and the high performance of the proposed algorithm.