More about HKUST
A Survey of 4D Reconstruction and Rendering
PhD Qualifying Examination
Title: "A Survey of 4D Reconstruction and Rendering"
by
Mr. Xuezhen WANG
Abstract:
Recent advancements in computer vision and graphics have shifted the focus
from static 3D reconstruction to the complex challenge of spatio-temporal (4D)
modeling. This survey explores the evolution of reconstruction techniques,
beginning with fundamental vision cues, including camera pose estimation,
dynamic points (coarse geometry) reconstruction, and spatio-temporal
registration, which serve as the structural foundation for understanding
moving scenes. We argue that the ultimate objective of these representations
is the simultaneous achievement of Novel View Synthesis (NVS) across both
spatial and temporal domains. By synthesizing seamless transitions in
viewpoint and time while maintaining rigorous geometric consistency, these
technologies enable breakthroughs in autonomous systems and immersive media.
This survey categorizes current methodologies, discusses the transition from
optimization-based methods to feed-forward methods, and identifies potential
future directions in achieving unified, real-time 4D scene reconstruction and
rendering.
Date: Thursday, 24 April 2026
Time: 5:30pm - 7:30pm
Venue: Room 2132C
Lift 22
Committee Members: Prof. Pedro Sander (Supervisor)
Dr. Dan Xu (Chairperson)
Prof. Long Quan