Micro-macro Wavelet-based Gaussian Splatting for 3D Reconstruction from Unconstrained Images

TL;DR

This paper introduces MW-GS, a novel 3D reconstruction method that disentangles scene features into multiple components using wavelet-based sampling and hierarchical fusion, achieving state-of-the-art results from unconstrained images.

Contribution

The paper proposes MW-GS, a new approach combining wavelet-based sampling and hierarchical residual fusion to improve 3D reconstruction from unconstrained images.

Findings

Achieves state-of-the-art rendering performance.

Effectively disentangles scene representations into global, refined, and intrinsic components.

Outperforms existing methods in 3D reconstruction quality.

Abstract

3D reconstruction from unconstrained image collections presents substantial challenges due to varying appearances and transient occlusions. In this paper, we introduce Micro-macro Wavelet-based Gaussian Splatting (MW-GS), a novel approach designed to enhance 3D reconstruction by disentangling scene representations into global, refined, and intrinsic components. The proposed method features two key innovations: Micro-macro Projection, which allows Gaussian points to capture details from feature maps across multiple scales with enhanced diversity; and Wavelet-based Sampling, which leverages frequency domain information to refine feature representations and significantly improve the modeling of scene appearances. Additionally, we incorporate a Hierarchical Residual Fusion Network to seamlessly integrate these features. Extensive experiments demonstrate that MW-GS delivers state-of-the-art rendering performance, surpassing existing methods.