Low-Rank Approximation for Efficient Compression of Gaussian Splatting Spherical Harmonics
A systematic study that reveals and leverages the intrinsic low-rank structure of SH coefficients in 3DGS for efficient compression.
I am currently a PhD candidate at the College of Information Science and Electronic Engineering of Zhejiang University (ZJU) (2021-Present), supervised by Prof. Lu Yu, with guidance from Asst. Prof. Yiyi Liao.
I got my B.Eng. degree in Electronic Information Engineering at the Qiushi Honors College (now known as the College of Future Technology) of Tianjin University (TJU) in 2021.
My research interests lie in 3D computer vision and visual data compression.
A systematic study that reveals and leverages the intrinsic low-rank structure of SH coefficients in 3DGS for efficient compression.
A GPU-accelerated Dual-Hilbert scan algorithm mapping 3D Gaussians to 2D videos, enabling efficient 3DGS compression over 500× faster than PLAS.
A lightweight preprocessing method canonicalizing scale and rotation in 3DGS to reduce redundancy.
A highly modular framework for compressing static/dynamic Gaussian splats, supporting video and point cloud codecs.
A neural mesh refinement method that utilizes a learned geometric prior on fine shapes to adaptively refine coarse meshes.
An end-to-end learnt point cloud coding framework for 3D machine intelligent tasks with BEV representation.
My nickname is "繁星知微" (Shine like stella, Perceive the subtle).
I like all kinds of sports, including running, cycling, basketball, badminton, frisbee and so on.
I also like photography, the magic of freezing time. I love using my lens to capture the beauty.
Welcome to follow my social media accounts: