Research map / 3D vision / Depth and stereo
Robust Pseudo Random Fields for Light-Field Stereo Matching
Chao-Tsung Huang
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 |
|---|---|---|---|
| use Reciprocal energy model instead of Gaussian for view-wise difference fit (ablation) | RPRF · StillLife view-wise difference | KL divergence: 0.1573 -> 0.1238 | Improved |
| use Reciprocal energy model instead of Gaussian for spatial difference fit (ablation) | RPRF · StillLife spatial difference | KL divergence: 0.1493 -> 0.0838 | Improved |
| use Reciprocal energy model instead of Gaussian for spatial difference fit (ablation) | RPRF · Medieval spatial difference | KL divergence: 0.1929 -> 0.0883 | Improved |
| use Reciprocal energy model instead of Gaussian for view-wise difference fit (ablation) | RPRF · Medieval view-wise difference | KL divergence: 0.0668 -> 0.0515 | Improved |
| use robust hard-EM energy for inference instead of soft-EM energy (ablation) | RPRF · StillLife depth maps | depth edge quality (qualitative) | Improved |
| initialize with over-smooth disparity (lambda=600) rather than default (ablation) | RPRF · StillLife 3x3 | DMSE after one update: 19.69 -> 1.21 | No clear effect |
| initialize with default parameters (lambda=300) rather than ground-truth depth (ablation) | RPRF · StillLife 3x3 | DMSE after one update: 4.76 -> 1.22 | No clear effect |
| add a second parameter-depth iteration (ablation) | RPRF · HCI Blender 9x9 | DMSE: 0.2% | Improved |
| apply RPRF to BM3D-denoised light fields instead of noisy ones (ablation) | RPRF · Noisy light fields | DMSE | Improved |
| reduce number of views from dense to sparse (denoised light fields) (ablation) | RPRF · Denoised light fields | DMSE | Hurt |
| use fewer views (sparse 3x3 instead of 9x9) (ablation) | RPRF · Clean light fields | DMSE | No clear effect |
Closest papers by meaning
- Empirical Bayesian Light-Field Stereo Matching by Robust Pseudo Random Field Modeling (TPAMI 2018 · similarity 0.98)
- Towards Multimodal Depth Estimation From Light Fields (CVPR 2022 · similarity 0.88)
- Digging Into Uncertainty-Based Pseudo-Label for Robust Stereo Matching (TPAMI 2023 · similarity 0.88)
- Hyperspectral Light Field Stereo Matching (TPAMI 2018 · similarity 0.88)
- Robust Light Field Depth Estimation for Noisy Scene With Occlusion (CVPR 2016 · similarity 0.87)
- Neural Markov Random Field for Stereo Matching (CVPR 2024 · similarity 0.86)
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.
