Publicado em: 27/01/2010

Defesa de Tese de Doutorado em Sistemas de Computação, dia 29/01, de Rafael Piccin Torchelsen

UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL
INSTITUTO DE INFORMÁTICA
PROGRAMA DE POS-GRADUAÇÃO EM COMPUTAÇÃO
———————————————————
DEFESA DE TESE DE DOUTORADO

Aluno: Rafael Piccin Torchelsen
Orientador: Prof. Dr. João Luiz Dihl Comba
Co-orientador: Prof. Dr. Rui Manuel Ribeiro de  Bastos (NVIDIA)

Titulo: Using the parametric domain for efficient computation

Área de Pesquisa: Sistemas de Computação

Data: 29/01/2010
Hora: 14:00h
Local: Auditório Prédio 67 – Andar Inferior

Banca Examinadora:
Profa. Dra. Soraia Raupp Musse (PUCRS)
Prof. Dr. Luis Gustavo Nonato (USP – São Carlos)
Prof. Dr. Marcelo Walter (UFPE)
Profa. Dra. Luciana Porcher Nedel (UFRGS)

Presidente da Banca: Prof. Dr. João Luiz Dihl Comba

Resumo:
The process of parameterizing a mesh to the plane is an ongoing research topic. Although there are several works dedicated to parameterization techniques the use of the resulting parameterizations has received less attention.
This work presents contributions related to the use of the parametric space to improve the computational efficiency of several algorithms. The main motivation comes from the fact that some algorithms are more efficiently computed on the parametric version of the mesh, compared to the 3-D version. For example, shortest distances can be computed, usually, an order of magnitude faster on the parametric space.
The contributions of this work can be applied to at least three research fields related to computer graphics: displacement mapping, distance computation on the surface of triangular meshes and agent path planning.
The contribution related to displacement mapping, presented in chapter 4, is used to increase the rendering performance and visual quality of terrains in games. The new method to compute distances, presented in chapter 5, increases the efficiency of several distance computation algorithms. This new method was also used on a novel agent path planning algorithm, to navigate agents on the surface of arbitrary meshes. This technique is presented in chapter 6.
The potential of the new distance computation method is not restricted to the applications presented in this thesis. In general, any technique that uses distance computation on the surface of triangular meshes can have the performance improved by the method. We can cite the following applications: procedural texture generation, surface labeling, re-meshing, mesh segmentation, etc.

Palavras-Chave:  geometric meshes,  triangular meshes, parameterization, distance computation, motion planning.