VSE Seminar: Decomposition, Approximation and Reconstruction of Large Geometric Data

Jyh-Ming Lien
Assistant Professor, Department of Computer Science, GMU

Wednesday, Oct 31
10:30 - 11:30 AM
Research Hall, Room 163


Abstract:
Complex geometric data comprising millions to billions of elements are omnipresent in variety of domains. Examples include point clouds of urban landscapes, polygon soup representing buildings in major cities, and combinatorial data structures induced from the motion of a robot. The unstructured nature of the data and significant amount of noise pose grand challenges in designing efficient and practical geometric algorithms.

In this talk, I will provide an overview of these challenges and algorithmic solutions developed by me and my students for representing, and manipulating massive geometric data of shape and motion. In particular, I will focus the discussion on two of our main contributions: (1) approximate representations via decomposition and (2) robust mesh reconstruction and repair. I will also discuss their applications in the areas of CAD, GIS, and robotics. Our research results have attracted a wide range of interests from academia, open-source software communities and industry.

Bio:
Jyh-Ming Lien is an Assistant Professor in the Department of Computer Science. He directs the Motion and Shape Computing (MASC) group and is affiliated with the Autonomous Robotics Laboratory at George Mason University. He received his Ph.D. in Computer Science from Texas A&M University in 2006. Prior to joining George Mason in 2007, he was a postdoctoral researcher at UC Berkeley. His recent work focuses on shape decomposition, approximation and reconstruction of complex and dynamic 3D geometries. His research has been supported by NSF, USGS, DOT, AFOSR, and Virginia Center for Innovative Technology. More images, videos, papers, and software about his work can be found at: https://masc.cs.gmu.edu/