Research map / 3D vision / Depth and stereo
Rational Polynomial Camera Model Warping for Deep Learning Based Satellite Multi-View Stereo Matching
Jian Gao, Jin Liu, Shunping Ji
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace fitted homography warping with rigorous RPC warping (ablation) | RED-Net · TLC SatMVS (2048x1472 patches) | MAE (m): 2.171 -> 1.945 | Improved |
| replace fitted homography warping with rigorous RPC warping (ablation) | CasMVSNet · TLC SatMVS (2048x1472 patches) | RMSE (m): 4.351 -> 3.841 | Improved |
| replace fitted homography warping with rigorous RPC warping (ablation) | UCS-Net · TLC SatMVS (2048x1472 patches) | MAE (m): 2.039 -> 2.026 | No clear effect |
| replace fitted homography warping with rigorous RPC warping (ablation) | CasMVSNet · TLC SatMVS (2048x1472 patches) | Completeness (%): 82.33 -> 81.54 | Hurt |
| replace fitted homography warping with rigorous RPC warping (ablation) | RED-Net · TLC SatMVS (5120x5120 full-size) | MAE (m): 2.517 -> 1.946 | Improved |
| replace fitted homography warping with rigorous RPC warping (ablation) | RED-Net · TLC SatMVS (5120x5120 full-size) | <2.5m (%): 66.42 -> 77.88 | Improved |
| process full 5120x5120 images instead of 2048x1472 patches | RED-Net (fitted homography) · TLC SatMVS | MAE (m): 2.171 -> 2.517 | Hurt |
| process full 5120x5120 images instead of 2048x1472 patches | SatMVS(RED-Net) · TLC SatMVS | MAE (m): 1.945 -> 1.946 | No clear effect |
| use RPC warping instead of homography warping, runtime (ablation) | RED-Net · TLC SatMVS (2048x1472 patches) | Runtime: 9min15s -> 13min52s | Hurt |
| use learning-based MVS (SatMVS RED-Net) instead of adapted COLMAP | adapted COLMAP · TLC SatMVS (2048x1472 patches) | RMSE (m): 5.291 -> 4.070 | Improved |
| increase image patch size for pin-hole fitting from 768x768 to 23040x23040 | RPC-to-pinhole fitting · Fitting error | Max fitting error (pixels): 0.16204 -> 6.62046 | Hurt |
Closest papers by meaning
- A Novel Recurrent Encoder-Decoder Structure for Large-Scale Multi-View Stereo Reconstruction From an Open Aerial Dataset (CVPR 2020 · similarity 0.86)
- Georeferenced Cross-View 3D Reconstruction from Ground and Satellite Images (NeurIPS 2026 · similarity 0.84)
- RPC-GS: Gaussian Splatting with native RPC Rendering for Satellite Imagery (NeurIPS 2026 · similarity 0.84)
- SD-MVS: Segmentation-Driven Deformation Multi-View Stereo with Spherical Refinement and EM Optimization (AAAI 2024 · similarity 0.83)
- Multi-view Reconstruction via SfM-guided Monocular Depth Estimation (CVPR 2025 · similarity 0.82)
- RRT-MVS: Recurrent Regularization Transformer for Multi-View Stereo (AAAI 2025 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
