Problem
Research questions and friction points this paper is trying to address.
motion blur
sparse views
3D reconstruction
free-viewpoint rendering
camera pose-free
CasDeblurGS: Cascaded 2D-to-3D Multi-View Consistency for 3D Gaussian Splatting from Two Blurry Images
🤖 AI Summary
This work proposes a cascaded 2D-to-3D reconstruction framework that, for the first time, achieves highly consistent 3D Gaussian Splatting from only two motion-blurred images—without requiring known camera poses, auxiliary sharp images, or per-scene optimization. The method establishes multi-view consistency through occlusion-aware filtering of local correspondences and introduces a pose-free 3D Gaussian representation to provide dense re-rendering guidance, enabling joint optimization of deblurring and novel view synthesis. Evaluated on both real and synthetic Deblur-NeRF datasets, the approach outperforms strong baselines by significant margins, achieving PSNR improvements of 1.19 dB and 2.11 dB, respectively, while simultaneously enhancing rendering fidelity and geometric consistency.
📝 Abstract
Free-viewpoint 3D scene media is increasingly important for immersive applications, yet practical capture often suffers from severe view sparsity and motion blur. Although neural rendering has advanced sparse-view synthesis, existing blur-aware methods typically require substantial multi-view redundancy, accurate camera poses, or costly per-scene optimization. We address a stringent yet practical setting: reconstructing a coherent 3D scene from only two motion-blurred images with known intrinsics, without input-view poses, auxiliary sharp images, or per-scene test-time optimization. To this end, we propose CasDeblurGS, a cascaded framework that progressively recovers reliable cross-view information from local 2D correspondences to global 3D guidance. Stage 1 constructs locally reliable guidance through occlusion-aware correspondence filtering, while Stage 2 aggregates the intermediate restorations into a provisional pose-free 3D Gaussian representation whose input-view re-renders provide dense global guidance for final restoration. The resulting views enable a more coherent 3D representation and higher-quality novel-view synthesis. Experiments on real-world and synthetic Deblur-NeRF scenes show consistent gains over strong baselines, improving PSNR by 1.19 dB and 2.11 dB, respectively. Progressive ablations, cross-view correspondence visualization, and camera reprojection analysis further demonstrate improvements in both rendering quality and multi-view geometric consistency.
Innovation
Methods, ideas, or system contributions that make the work stand out.
3D Gaussian Splatting
motion deblurring
sparse-view reconstruction
multi-view consistency
pose-free rendering
🔎 Similar Papers
No similar papers found.
