InFusionSurf: Refining Neural RGB-D Surface Reconstruction Using Per-Frame Intrinsic Refinement and TSDF Fusion Prior Learning
This work addresses camera motion-induced blurs and reconstruction details for applications in 3D modeling and computer vision, representing an incremental enhancement over existing methods.
The paper tackles the problem of improving 3D surface reconstruction quality from RGB-D video frames in neural radiance field frameworks by refining depth information and using TSDF Fusion as a pretraining tool, resulting in high accuracy while maintaining optimization efficiency.
We introduce InFusionSurf, an innovative enhancement for neural radiance field (NeRF) frameworks in 3D surface reconstruction using RGB-D video frames. Building upon previous methods that have employed feature encoding to improve optimization speed, we further improve the reconstruction quality with minimal impact on optimization time by refining depth information. InFusionSurf addresses camera motion-induced blurs in each depth frame through a per-frame intrinsic refinement scheme. It incorporates the truncated signed distance field (TSDF) Fusion, a classical real-time 3D surface reconstruction method, as a pretraining tool for the feature grid, enhancing reconstruction details and training speed. Comparative quantitative and qualitative analyses show that InFusionSurf reconstructs scenes with high accuracy while maintaining optimization efficiency. The effectiveness of our intrinsic refinement and TSDF Fusion-based pretraining is further validated through an ablation study.