I4-5 - Sedma Nacionalna Konferencija so Me|unarodno U~estvo...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
EXPERIMENTS IN WAVELET PROCESSING OF 3D OBJECTS FOR SIMILARITY COMPARISON Celakovski Sasko 1 , Danco Davcev 2 , Vladimir Trajkovik 3 1 Cosmofon – Skopje, sasko.celakovski@cosmofon.com.mk 2 Faculty of Electrical Engineering – Skopje, etfdav@etf.ukim.edu.mk 3 Faculty of Electrical Engineering – Skopje, trvlado@etf.ukim.edu.mk Abstract - In this study, we purpose a new method for similarity comparison of 3D objects based on wavelet transformation on the 3D data. The wavelet filtering generates a set of coefficients (details) that express the spatial-frequency characteristics of the analyzed 3D object. The extracted wavelet coefficients are stored in a segmented feature vector to describe the 3D object in various resolution levels. We have experimentally shown that this feature vector can be successfully used in comparison of 3D objects. Keywords - Comparison of 3D objects, 3D Wavelet Transform, Multiresolution processing. 1. INTRODUCTION Previous work on the problems of feature extraction and similarity comparison of 3D objects presents different algorithms for description of 3D objects and design of comparison systems that exploit their advantages. Dibio [1] proposes a complete solution to the problem of 3D recognition using shape information extracted from range images and parameterized volumetric models. Geometric modeling of the data with parametric surfaces was presented as a method for archiving and searching 3D objects by Razdan et. al. [2]. System for description and similarity comparison of 3D objects aided by active learning was presented by Cha [3]. Cha’s system uses volume-surface ratio, moment invariant and Fourier transform coefficients [4] for feature representation of the 3D objects. Vranic et. al. [5] presents a method for description of 3D objects based on feature extraction using Fourier Transform of a voxelized 3D object. In [5] the absolute values of the complex coefficients from the Fourier transform are considered as components of the feature vector. In this study, we present a new method for similarity comparison of 3D objects based on feature extraction from 3D models. The proposed algorithm is based on wavelet transform (WT) and multiresolution representation (MRR) of the polygonal 3D object. The feature vector is built from the wavelet coefficient at different resolution levels of the 3D object. 2. FEATURE EXTRACTION The goal of feature extraction algorithm is to define 3D shape descriptor that describes the 3D object in a way that can be used in determination of similarity with other 3D objects. It should enable similarity comparisons of generated descriptors in terms of “close” points in the feature vector. Before the feature extraction is performed the 3D object should be transformed into a canonical space. This is a normalization step that will eliminate the
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

Page1 / 5

I4-5 - Sedma Nacionalna Konferencija so Me|unarodno U~estvo...

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online