Unblur-SLAM: Dense Neural SLAM for Blurry Inputs

2026-03-26

AI Summary

This work addresses the challenge of achieving high-precision pose estimation and sharp 3D reconstruction in RGB SLAM under motion blur and defocus blur, conditions under which conventional SLAM systems typically degrade. The paper presents the first RGB SLAM framework that deeply integrates adaptive deblurring with neural SLAM. For frames amenable to effective deblurring, it refines poses and depth through multi-view optimization; for frames where deblurring fails, it jointly models the blur formation process and sub-frame poses to collaboratively recover sharp scene details. The method synergistically combines a feedforward deblurring network, local–global optimization, loop closure detection, and 3D Gaussian Splatting, enabling robust handling of diverse blur types with dynamically adjusted processing strategies. Experiments demonstrate significant improvements in both pose accuracy and reconstruction sharpness across multiple real-world datasets, achieving state-of-the-art performance.

Abstract

We propose Unblur-SLAM, a novel RGB SLAM pipeline for sharp 3D reconstruction from blurred image inputs. In contrast to previous work, our approach is able to handle different types of blur and demonstrates state-of-the-art performance in the presence of both motion blur and defocus blur. Moreover, we adjust the computation effort with the amount of blur in the input image. As a first stage, our method uses a feed-forward image deblurring model for which we propose a suitable training scheme that can improve both tracking and mapping modules. Frames that are successfully deblurred by the feed-forward network obtain refined poses and depth through local-global multi-view optimization and loop closure. Frames that fail the first stage deblurring are directly modeled through the global 3DGS representation and an additional blur network to model multiple blurred sub-frames and simulate the blur formation process in 3D space, thereby learning sharp details and refined sub-frame poses. Experiments on several real-world datasets demonstrate consistent improvements in both pose estimation and sharp reconstruction results of geometry and texture.

Problem

• blurry inputs
• 3D reconstruction
• SLAM
• motion blur
• defocus blur

Innovation

• Neural SLAM
• Image Deblurring
• 3D Gaussian Splatting
• Motion Blur
• Adaptive Computation