Locally and globally shape aware volumetric understanding

Manish Goyal, Purdue University

Abstract

The process of reconstructing CAD models from physical parts, formally known as digital shape reconstruction (DSR) is an integral part of product development. While, the majority of current methods used in DSR are surface based our overarching goal is to obtain parametrization of 3d meshes while retaining the volumetric information of the original part. As a first step to achieve this goal of volumetric understanding, we extract (1) locally prominent cross-sections (PCS) and (2) organize them into sweep components. For organizing PCS we introduce two new algorithms derived from Locally Linear Embedding (LLE) and Affinity Propagation(AP). We extract partial PCS (PPCS) in the region of sweep intersection, and full PCS (FPCS) elsewhere. The LLE based algorithm analyzes the (PCS) using their geometric properties to build global manifold. The AP based algorithm then clusters the local cross sections by propagating affinities among them in the embedded space to form different global volumetric sweep components. Our approach thus facilitates parametrization of a model into different sweep components by avoiding the actual segmentation of the mesh into different surfaces. We demonstrate the application of our method through the extraction of volumetric information of various CAD parts.

Degree

M.S.E.

Advisors

Ramani, Purdue University.

Subject Area

Mechanical engineering

COinS