Approximation of CAD Models Using Skeletonization

Abstract

Skeletonization is a method to extract shape information of a CAD model with lower dimension. The associate information requirements of the skeleton are different from application to application. For instance, surface reconstruction requires skeleton with detailed geometry such as distances from the surfaces to the skeleton. On the other hand, object recognition requires only skeleton in order to perform similarity comparison. Even though skeletonization has been studied extensively, most of the previous methods are sensitive to noise, time consuming, or restricted to specific 3D CAD models. A practical method for skeletonization from general 3D CAD models using Centroidal Graph is discussed in this paper. The skeletonization process begins with the decomposition of the 3D CAD part model into subvolumes or cells. The loci of Centroids are then computed for all cells in order to complete the skeletonization process.