Extracting volume distribution feature from 3D models

Wenjun Dai,Mingyong Pang,Gangshan Wu,Fuyan Zhang

DOI: https://doi.org/10.13245/j.hust.2005.s1.091

2005-01-01

Abstract:The key issue of 3D model retrieval is how to extract features from a given model.Distribution of volume along certain direction in 3D space is one of the most important features for a geometric object.In this paper,we propose an algorithm to extract volumetric characters from arbitrary models based on half-edge data structure.The algorithm first rotates a mesh model to coordinate-axis-aligned position by employing Principal Component Analysis (PCA) method,and then uses a set of parallel equidistant planes to slice the model into several segments along three principal axes,respectively.From each cross section of the model,a polygon set composed of one or more polygons can be obtained and the areas of the cross section can be computed on the fly.After normalizing these areas,we plot three volume distribution curves along corresponding principal axes.The curves give a shape descriptor of the model.Experiments show that our algorithm is efficient,stable,and can process 3D models with arbitrary geometry and topology.