Meshless Extraction of Closed Feature Lines Using Histogram Thresholding

Abstract

In reverse engineering, the reconstruction of a surface model from a point cloud requires the extraction of closed feature lines at the borders of the different surface patches. In this paper we propose a new algorithm to extract such closed polygonal feature lines, representing sharp or smooth edges, from a point cloud. Based on the variation of the normal vectors and a graph approach we extract the sharp edges, which are used to divide the point cloud in smooth regions. Smooth edges, such as fillets, are extracted for each smooth region separately using a novel approach for 1D-histogram thresholding: we use the curvature histogram in a multi-resolution manner in order to split a point set in different regions of similar curvature (patches). The polygonal smooth edges at the borders between these different patches are extracted by point clustering and processing a graph of the point clusters.

Feature Surface Extraction and Reconstruction from Industrial Components Using Multistep Segmentation and Optimization

“…However, this process requires considerable computing resources to segment the point cloud using normal vectors to define point features rather than point coordinates. However, we determine that computing resources will not be consumed in large quantities due to the small number of points in the industrial component point cloud compared with buildings and complex man-made objects [13]. Nevertheless, the issue that the proposed method requires a large amount of computing resources should be taken seriously.…”

Digitized 3D mesh segmentation based on curvature analysis

“…For example, these distributions can be used in image processing to improve compression [10]. Some methods in 3D mesh processing also begin to use distributions, for example to apply segmentation [4] or extract object edges [6].…”