Seamless simplification of multi-chart textured meshes with adaptively updated correspondence

Meshes obtained by 3D scanning or photogrammetry are usually very dense and multi-chart textured. Effectively reducing the amount of data while preserving both geometry and appearance is an important and also challenging problem. This paper presents a new method for seamless simplification of dense...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhang, Wenjing, Zheng, Jianmin, Cai, Yiyu, Ynnerman, Anders
Other Authors: School of Mechanical and Aerospace Engineering
Format: Article
Language:English
Published: 2022
Subjects:
Online Access:https://hdl.handle.net/10356/161834
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
Description
Summary:Meshes obtained by 3D scanning or photogrammetry are usually very dense and multi-chart textured. Effectively reducing the amount of data while preserving both geometry and appearance is an important and also challenging problem. This paper presents a new method for seamless simplification of dense triangular meshes with multi-chart textures. The method is composed of three components: geometry simplification, correspondence map construction and texture image generation. The core of the method lies in the ideas of decoupling geometry, UV-parameterization and texture processes to make the simplification process less constrained by the underlying attribute encoding, a reliable updating procedure for constructing the correspondence map between the simplified and original meshes, and a least squares model for filling texture elements, which are the key to producing artifact-free visual appearance. Compared to the state-of-the-art, the proposed method has several advantages: (1) it enables aggressive simplification with well preserved geometry and appearance; (2) it automatically achieves seamless simplification without special treatment on the seams between multi-charts of the texture; and (3) it is reliable. Experimental results have demonstrated these features.