Rethinking 3D Geometry Compression Through the Lens of Structural Representation
#3d
#art
#data-compression
#data-processing
#digital-twin
#artificial-intelligence
#machine-learning
#processing
Abstract:
With the rapid development of 3D applications in VR/AR, robotics, digital twins, and immersive communication, effective 3D geometry data compression has become more critical than ever. However, the inherently irregular and unstructured nature of raw 3D data poses significant technical challenges for traditional compression frameworks, which typically thrive on highly regular grids. This talk explores advanced 3D geometry representations through the lens of spatial structurization across 1D, 2D, and 3D spaces. Specifically, we will examine how mapping complex, irregular geometric structures into structured 1D sequences, 2D image grids, or regular 3D volumetric data can successfully bridge the gap between raw spatial data and highly optimized, classical compression pipelines. By rethinking data representation as a fundamental prerequisite for efficient encoding, this presentation highlights novel, robust pathways toward ultra-efficient 3D data storage, transmission, and streaming, ultimately offering fresh insights into the future of geometric data processing.
Bio:
Dr. Junhui Hou is a Professor with the Department of Computer Science, City University of Hong Kong (CityUHK). His research interests include multidimensional visual computing, such as light field, hyperspectral, geometry, and event data. He received the Early Career Award and Research Fellow from the Hong Kong Research Grants Council, the Excellent Young Scientists Fund from NSFC, the IEEE TIP Best Paper Award, and the CityUHK Presidential Research Excellence Award for Junior Faculty. He is serving as a Senior Area Editor for IEEE TIP and an Associate Editor for IEEE TVCG and TMM, and served as an Associate Editor for IEEE TIP and TCSVT.
Date and Time
Location
Hosts
Registration
-
Add Event to Calendar
Loading virtual attendance info...