Digitized 3D mesh segmentation based on curvature analysis

Abstract

To cite this version:Silvère AbstractToday, it is increasingly frequent and easy to digitize the surface of real 3D objects. However, the obtained meshes are often inaccurate and noisy. In this paper, we present a method to segment a digitized 3D surface from a real object by analyzing its curvature. Experimental results -applied on real digitized 3D meshes -show the efficiency of our proposed analysis, particularly in a reverse engineering process.

Towards Automatic CAD Modeling from 3D Scan Sketch based Representation

“…Rule-based learning methods (Gauthier et al, 2019) focus on the recognition of two geometric features and their repetition patterns: counterboared and countersunk holes which require recognition of parallelism and concentricity relations. The recgonition rules are defined based on geometrical relations between primitives.…”

“…These methods can be classified into constrained primitive fitting (Kaiser et al, 2019;Patil et al, 2017;Kovács et al, 2015) and learning-based primitive fitting (Li et al, 2019;Ranftl and Koltun, 2018;Brachmann et al, 2017). There are also recent learning-based approaches that look for the similarity between scans and CADs to finally retrieve the appropriate CAD model (Dahnert et al, 2019;Avetisyan et al, 2019) or look for specific feature or repetition patterns (Gauthier et al, 2019).…”

Аналіз Щільності Полігональних Моделей У Тривимірних Комп’ютерних Іграх

“…To preserve the properties of complex curved surfaces, an increase in the number of surface triangles is necessary. The curvature of a three-dimensional surface at a point can be represented by Gaussian [7] or mean [7] curvature. The Gaussian curvature of the surface K is calculated [8] as the product of the principal curvatures 1 2 , k k (the maximum and minimum curvatures of the object's surface intersections with the normal planes)…”