Splat-based 3D Scene Reconstruction with Extreme Motion-blur

2026-07-18

AI Summary

This work addresses the severe motion blur in RGB-D images caused by long exposures under low-light conditions, which compromises pose estimation and degrades 3D reconstruction quality. The paper introduces Gaussian splatting into extreme motion blur scenarios for the first time and proposes a joint optimization framework that simultaneously recovers sharp images, refines camera poses, and optimizes geometry. By integrating optical flow and ICP to align RGB-D frames and modeling motion blur through continuous camera trajectories during exposure, the method achieves effective deblurring and depth alignment. Evaluated on a newly constructed extreme-blur RGB-D dataset, the approach significantly outperforms existing methods, enabling high-quality 3D reconstruction for applications in robotics, autonomous driving, and augmented reality.

Abstract

We propose a splat-based 3D scene reconstruction method from RGB-D input that effectively handles extreme motion blur, a frequent challenge in low-light environments. Under dim illumination, RGB frames often suffer from severe motion blur due to extended exposure times, causing traditional camera pose estimation methods, such as COLMAP, to fail. This results in inaccurate camera pose and blurry color input, compromising the quality of 3D reconstructions. Although recent 3D reconstruction techniques like Neural Radiance Fields and Gaussian Splatting have demonstrated impressive results, they rely on accurate camera trajectory estimation, which becomes challenging under fast motion or poor lighting conditions. Furthermore, rapid camera movement and the limited field of view of depth sensors reduce point cloud overlap, limiting the effectiveness of pose estimation with the ICP algorithm. To address these issues, we introduce a method that combines camera pose estimation and image deblurring using a Gaussian Splatting framework, leveraging both 3D Gaussian splats and depth inputs for enhanced scene representation. Our method first aligns consecutive RGB-D frames through optical flow and ICP, then refines camera poses and 3D geometry by adjusting Gaussian positions for optimal depth alignment. To handle motion blur, we model camera movement during exposure and deblur images by comparing the input with a series of sharp, rendered frames. Experiments on a new RGB-D dataset with extreme motion blur show that our method outperforms existing approaches, enabling high-quality reconstructions even in challenging conditions. This approach has broad implications for 3D mapping applications in robotics, autonomous navigation, and augmented reality. Both code and dataset are publicly available on https://github.com/KAIST-VCLAB/gs-extreme-motion-blur.

Problem

• motion blur
• 3D scene reconstruction
• camera pose estimation
• low-light conditions
• RGB-D

Innovation

• Gaussian Splatting
• motion blur
• RGB-D reconstruction
• camera pose estimation
• image deblurring