Large Images are Gaussians: High-Quality Large Image Representation with Levels of 2D Gaussian Splatting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhu, Lingting, Lin, Guying, Chen, Jinnan, Zhang, Xinjie, Jin, Zhenchao, Wang, Zhao, Yu, Lequan
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913689281495040
author Zhu, Lingting
Lin, Guying
Chen, Jinnan
Zhang, Xinjie
Jin, Zhenchao
Wang, Zhao
Yu, Lequan
author_facet Zhu, Lingting
Lin, Guying
Chen, Jinnan
Zhang, Xinjie
Jin, Zhenchao
Wang, Zhao
Yu, Lequan
contents While Implicit Neural Representations (INRs) have demonstrated significant success in image representation, they are often hindered by large training memory and slow decoding speed. Recently, Gaussian Splatting (GS) has emerged as a promising solution in 3D reconstruction due to its high-quality novel view synthesis and rapid rendering capabilities, positioning it as a valuable tool for a broad spectrum of applications. In particular, a GS-based representation, 2DGS, has shown potential for image fitting. In our work, we present \textbf{L}arge \textbf{I}mages are \textbf{G}aussians (\textbf{LIG}), which delves deeper into the application of 2DGS for image representations, addressing the challenge of fitting large images with 2DGS in the situation of numerous Gaussian points, through two distinct modifications: 1) we adopt a variant of representation and optimization strategy, facilitating the fitting of a large number of Gaussian points; 2) we propose a Level-of-Gaussian approach for reconstructing both coarse low-frequency initialization and fine high-frequency details. Consequently, we successfully represent large images as Gaussian points and achieve high-quality large image representation, demonstrating its efficacy across various types of large images. Code is available at {\href{https://github.com/HKU-MedAI/LIG}{https://github.com/HKU-MedAI/LIG}}.
format Preprint
id arxiv_https___arxiv_org_abs_2502_09039
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Large Images are Gaussians: High-Quality Large Image Representation with Levels of 2D Gaussian Splatting
Zhu, Lingting
Lin, Guying
Chen, Jinnan
Zhang, Xinjie
Jin, Zhenchao
Wang, Zhao
Yu, Lequan
Computer Vision and Pattern Recognition
Artificial Intelligence
While Implicit Neural Representations (INRs) have demonstrated significant success in image representation, they are often hindered by large training memory and slow decoding speed. Recently, Gaussian Splatting (GS) has emerged as a promising solution in 3D reconstruction due to its high-quality novel view synthesis and rapid rendering capabilities, positioning it as a valuable tool for a broad spectrum of applications. In particular, a GS-based representation, 2DGS, has shown potential for image fitting. In our work, we present \textbf{L}arge \textbf{I}mages are \textbf{G}aussians (\textbf{LIG}), which delves deeper into the application of 2DGS for image representations, addressing the challenge of fitting large images with 2DGS in the situation of numerous Gaussian points, through two distinct modifications: 1) we adopt a variant of representation and optimization strategy, facilitating the fitting of a large number of Gaussian points; 2) we propose a Level-of-Gaussian approach for reconstructing both coarse low-frequency initialization and fine high-frequency details. Consequently, we successfully represent large images as Gaussian points and achieve high-quality large image representation, demonstrating its efficacy across various types of large images. Code is available at {\href{https://github.com/HKU-MedAI/LIG}{https://github.com/HKU-MedAI/LIG}}.
title Large Images are Gaussians: High-Quality Large Image Representation with Levels of 2D Gaussian Splatting
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2502.09039