An Algorithm for Extracting Edge Feature Points of Point Clouds based on Plane Segmentation and Dimensionality Reduction

Edge feature extraction in 3D point cloud constitutes an essential component of point cloud processing and reconstruction. Addressing the issue of inaccurate extraction of the feature point set for the egde of building point cloud, this paper presents a point cloud edge extraction approach based on the random sample consensus algorithm and the dimension reduction α-shape concave bag algorithm. The methodology encompasses two steps: plane extraction and edge feature extraction. In plane extraction, the random sample consensus algorithm is employed to extract the point cloud plane. In edge feature extraction, in accordance with the distribution characteristics of planar point cloud in three-dimensional space, Principal Component Analysis (PCA) is utilized to reduce the dimension of the planar point cloud, thereby obtaining the two-dimensional principal component point set. The two-dimensional edge feature point set is acquired by computing the concave envelope boundary of the two-dimensional principal component point set, and subsequently, the two-dimensional feature point set is mapped to three-dimensional feature points. The experimental outcomes demonstrate that the proposed method can effectively segregate and extract the edge contour point cloud of the building point cloud, and exhibits high accuracy and efficiency.