DenoiseGS: Gaussian Reconstruction Model for Burst Denoising

2025-11-28

AI Summary

Existing burst denoising methods struggle to simultaneously ensure robustness to large motions and computational efficiency, often causing geometric degradation of 3D Gaussian point clouds and loss of high-frequency details. This paper proposes the first efficient 3D Gaussian point lattice denoising framework tailored for handheld burst imaging. We innovatively introduce a Gaussian self-consistency loss (GSC) and a logarithmically weighted frequency-domain loss (LWF), enabling geometric regularization via model-self-generated high-quality point clouds and enhancing high-frequency detail recovery while avoiding domain shift. Our method integrates 3D Gaussian reconstruction, a lightweight feed-forward network, and self-supervised learning. Experiments demonstrate that our approach significantly outperforms NeRF-based baselines on both burst denoising and novel-view synthesis under noise, achieving a 250× speedup in inference time. The framework enables high-fidelity, real-time-capable 3D reconstruction.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Burst denoising methods are crucial for enhancing images captured on handheld devices, but they often struggle with large motion or suffer from prohibitive computational costs. In this paper, we propose DenoiseGS, the first framework to leverage the efficiency of 3D Gaussian Splatting for burst denoising. Our approach addresses two key challenges when applying feedforward Gaussian reconsturction model to noisy inputs: the degradation of Gaussian point clouds and the loss of fine details. To this end, we propose a Gaussian self-consistency (GSC) loss, which regularizes the geometry predicted from noisy inputs with high-quality Gaussian point clouds. These point clouds are generated from clean inputs by the same model that we are training, thereby alleviating potential bias or domain gaps. Additionally, we introduce a log-weighted frequency (LWF) loss to strengthen supervision within the spectral domain, effectively preserving fine-grained details. The LWF loss adaptively weights frequency discrepancies in a logarithmic manner, emphasizing challenging high-frequency details. Extensive experiments demonstrate that DenoiseGS significantly exceeds the state-of-the-art NeRF-based methods on both burst denoising and novel view synthesis under noisy conditions, while achieving extbf{250$ imes$} faster inference speed. Code and models are released at https://github.com/yscheng04/DenoiseGS.

Problem

• Enhances burst image denoising for handheld devices under large motion
• Addresses degradation of Gaussian point clouds from noisy input images
• Preserves fine details in images through spectral domain supervision

Innovation

• 3D Gaussian Splatting for burst denoising
• Gaussian self-consistency loss for geometry regularization
• Log-weighted frequency loss for detail preservation