2D-LLFE-NVC: 2D Laser Line Feature Extraction Using Normal Vector Clustering
Jinshun Ou, Jiying Ren, Wenguang Li, Panling Huang, Jun Zhou, Jiehan Zhou, Tao Zhang · 2024
This paper presents a novel 2D laser data line feature extraction algorithm based on normal vector clustering(2D-LLFE-NVC). The algorithm consists of normal vector calculation, normal normalization based on multi-scale operator normal difference and median filter, and line feature extraction based on normal vector clustering. Unlike existing methods, 2D-LLFE-NVC leverages the continuity of laser point clouds and detects breakpoints using multi-scale operators based on normal differences. The extracted line features are more accurate and correct on a global scale. To validate its performance, we conducted SLAM experiments in a 5000-square-meter indoor environment using both the proposed algorithm and existing methods. By comparing the line feature points extracted by these different approaches with manually annotated line feature points, and demonstrating the superior performance of the proposed method in terms of correctness and completeness.