Research map / Probabilistic and scientific modeling / Molecules and proteins
Bridging Local–Global Dissonance: Learning from Compressive Measurements for Hyperspectral Reconstruction
Xian-Hua Han
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 heterogeneous HSRA token mixing instead of all-ConvFormer (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.40 -> 36.08 | Improved |
| use heterogeneous HSRA token mixing instead of all-MK-PTM (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.47 -> 36.08 | Improved |
| use heterogeneous HSRA token mixing instead of all-Swin (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 34.98 -> 36.08 | Improved |
| use heterogeneous HSRA token mixing instead of all-HM-SSA (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.11 -> 36.08 | Improved |
| raise MK-PTM channel ratio from r=0.5 to r=1 (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.90 -> 36.08 | Improved |
| use multi-kernel MK-PTM instead of 3x3 only (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.92 -> 36.08 | Improved |
| add LWIM (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.89 -> 36.08 | Improved |
| add large-window attention (2w) in HM-SSA (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | PSNR (dB): 35.65 -> 36.08 | Improved |
| add large-window attention (2w) in HM-SSA (ablation) | HSRA-S · KAIST simulation (ablation on HSRA-S) | ERGAS (lower better): 22.631 -> 21.542 | Improved |
| scale HSRA-S to HSRA-L | HSRA E2E · KAIST 10 simulation scenes (E2E) | PSNR (dB): 36.08 -> 37.49 | Improved |
| HSRA-L versus DWMT at lower compute | DWMT · KAIST 10 simulation scenes (E2E) | PSNR (dB): 36.82 -> 37.49 | Improved |
| embed HSRA in deep unfolding instead of SSR-3stg prior | SSR-3stg · KAIST 10 simulation scenes (deep unfolding) | PSNR (dB): 39.01 -> 39.40 | Improved |
Closest papers by meaning
- Uncertainty-Driven Spectral Compressive Imaging with Spatial-Frequency Transformer (ECCV 2024 · similarity 0.89)
- OneHSI: A Unified Hyperspectral Foundation Model with Physical Consistency (ECCV 2026 · similarity 0.89)
- Hyperspectral Image Reconstruction Using a Deep Spatial-Spectral Prior (CVPR 2019 · similarity 0.88)
- LESSViT: Robust Hyperspectral Representation Learning under Spectral Configuration Shift (NeurIPS 2026 · similarity 0.88)
- Detail Matters: Mamba-Inspired Joint Unfolding Network for Snapshot Spectral Compressive Imaging (AAAI 2025 · similarity 0.86)
- Sp3ctralMamba: Physics-Driven Joint State Space Model for Hyperspectral Image Reconstruction (AAAI 2025 · 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.
