High-quality Textured Surface Reconstruction from Registered Images: State-of-the-art Methods

PhD Qualifying Examination


Title: "High-quality Textured Surface Reconstruction from Registered
Images: State-of-the-art Methods"

by

Mr. Shiwei LI


Abstract:

With the emergence of a new era of 3D technology (e.g., 3D printing, 
virtual reality), capturing 3D data from reality will be a heated demand 
in future years. The rapid development of consumer camera has boosted the 
quality of image-based 3D reconstruction. Saving the 3D models of objects, 
human and buildings with a camera or even a mobile phone handily and 
instantly is not an imagination. In the coming decade, this versatile and 
low-cost solution is the main trend for many practical applications.

This survey first gives an overview of the entire image-based 3D 
reconstruction pipelines. Then it introduces existing works regarding to 
three stages in the pipeline, namely Multi-View Stereo (MVS), Surface 
Generation and Texture Mapping. Some state-of-the-art algorithms will be 
deeply reviewed. After the survey, an efficient surface refinement 
algorithm is presented, which is crucial to the final reconstruction 
quality. At the end, we discuss some of the unsolved problems in existing 
methods, and propose potential solutions as our future work.


Date:			Wednesday, 22 June 2016

Time:                  	2:00pm - 4:00pm

Venue:                  Room 2611
                         Lifts 31/32

Committee Members:	Prof. Long Quan (Supervisor)
 			Prof. Chiew-Lan Tai (Chairperson)
 			Prof. Ting-Chuen Pong
 			Prof. Huamin Qu


**** ALL are Welcome ****