Sat-DN: Implicit Surface Reconstruction from Multi-View Satellite Images with Depth and Normal Supervision

TL;DR

Sat-DN is a new framework for reconstructing detailed 3D terrain and building facades from multi-view satellite images, using multi-resolution hash grids, depth guidance, and normal constraints for faster and more accurate results.

Contribution

It introduces a multi-resolution hash grid architecture with progressive training, explicit depth and normal supervision, improving satellite image-based surface reconstruction.

Findings

Outperforms existing methods on DFC2019 dataset

Achieves state-of-the-art qualitative and quantitative results

Accelerates training with multi-resolution hash grid

Abstract

With advancements in satellite imaging technology, acquiring high-resolution multi-view satellite imagery has become increasingly accessible, enabling rapid and location-independent ground model reconstruction. However, traditional stereo matching methods struggle to capture fine details, and while neural radiance fields (NeRFs) achieve high-quality reconstructions, their training time is prohibitively long. Moreover, challenges such as low visibility of building facades, illumination and style differences between pixels, and weakly textured regions in satellite imagery further make it hard to reconstruct reasonable terrain geometry and detailed building facades. To address these issues, we propose Sat-DN, a novel framework leveraging a progressively trained multi-resolution hash grid reconstruction architecture with explicit depth guidance and surface normal consistency constraints to enhance reconstruction quality. The multi-resolution hash grid accelerates training, while the progressive strategy incrementally increases the learning frequency, using coarse low-frequency geometry to guide the reconstruction of fine high-frequency details. The depth and normal constraints ensure a clear building outline and correct planar distribution. Extensive experiments on the DFC2019 dataset demonstrate that Sat-DN outperforms existing methods, achieving state-of-the-art results in both qualitative and quantitative evaluations. The code is available at https://github.com/costune/SatDN.