Towards Accurate and Realistic Dense 3D Reconstruction

PhD Thesis Proposal Defence


Title: "Towards Accurate and Realistic Dense 3D Reconstruction"

by

Mr. Jingyang ZHANG


Abstract:

3D modeling is an important task to preserve and visualize real world scenes by 
computer. The applications include but not limited to heritage preserving, 
city-scale survey and AR/VR applications. Typical image-based 3D reconstruction 
methods include SfM, MVS, meshing and texturing. This thesis aims at improving 
the last three steps with neural network techniques so that the reconstruction 
pipeline is capable to produce 3D model with high accuracy and realistic 
appearance. First, we introduce visibility handling into MVS and mitigate the 
high memory consumption of previous learning-based MVS methods. Second, we 
investigate neural implicit surface reconstruction by geometry prior and 
differentiable rendering. Third, we optimize view-dependent appearance and 
surface material for reconstructed model. The proposed pipeline is extensively 
evaluated on multiple datasets, including both synthetic and real world data.


Date:			Wednesday, 2 March 2022

Time:                  	3:00pm - 5:00pm

Zoom Meeting:
https://hkust.zoom.us/j/92661414865?pwd=RkE5cGZMQzlMNXdpQ2xYQWdpajJMUT09

Committee Members:	Prof. Long Quan (Supervisor)
  			Prof. Chiew-Lan Tai (Chairperson)
 			Dr. Qifeng Chen
 			Prof. Pedro Sander


**** ALL are Welcome ****