- Rating
- 2
- Confidence
- 4
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 2
Read review
Paper summary
EquiPan addresses cross-sensor pansharpening through single-source training and fixed-weight deployment on unseen sensors. It combines paired spectral transformations, reversible spectral transport, and PAN mean alignment. Experiments across seven backbones and three unseen sensors show generally improved transfer over interpolation-only baselines.
Critical strengths
- The method addresses a practical problem: applying a trained model to another sensor without collecting new training labels or updating its weights.
- The framework works with several different fusion networks and generally improves their cross-sensor performance, suggesting that its benefits are not tied to a particular architecture.
Critical weaknesses
- I would have liked to see direct comparisons with closely related cross-sensor methods. Without these comparisons, I find it difficult to assess the advantage over existing approaches.
- The paper explains the method through approximate equivariance, but does not clearly establish what this contributes beyond paired data augmentation and statistical alignment.
Questions for authors
- How does EquiPan compare with closely related methods such as UniPAN under comparable conditions?
- What evidence shows that equivariance explains the improvement, beyond the use of paired spectral augmentation?
- What does the reversible mapping contribute beyond simpler spectral mapping and normalization?
- Since RST uses statistics from the target image, how does its deployment procedure differ from existing distribution-alignment methods?
3 comments
2. Evidence That Equivariance Explains Improvements Beyond Augmentation (W2 & Q2)
-
Manuscript Revision: As clarified in revised Sections 2, 3.1, and 3.2, conventional unilateral spectral augmentation breaks radiometric correspondence. In contrast, SCT applies bilaterally ( and ) around the fixed PAN spatial anchor. By the triangle inequality, minimizing the joint loss (Eq. 21) strictly upper-bounds the equivariance error .
-
New Ablations: We ablate: (i)
no_virtual(retaining RST+PMA alignment at test time but removing the virtual equivariance loss ), (ii)SCT Fixed 8(fixing virtual band count without 4-band sampling), (iii)SCT Selection(discrete band selection only, without continuous Gaussian response mixing ), and two non-equivariant physical augmentations: (iv)Gaussian_Blur(random blur on MS and GT) and (v)Pan_noise(random noise on PAN):
| Variant (ABFNet, WV3 Source) | WV3 PSNR / HQNR | WV2 PSNR / HQNR | QB PSNR / HQNR | GF2 PSNR / HQNR |
|---|---|---|---|---|
| EquiPan (Full) | 37.32 / 0.9528 | 31.85 / 0.9341 | 35.10 / 0.9121 | 44.00 / 0.9212 |
no_virtual (w/o ) | 37.47 / 0.9523 | 29.78 / 0.9327 | 33.50 / 0.8322 | 40.77 / 0.9298 |
SCT Fixed 8 ( only) | 37.40 / 0.9527 | 30.26 / 0.9334 | 34.43 / 0.7619 | 36.24 / 0.9337 |
SCT Selection ( only, ) | 37.33 / 0.9547 | 30.70 / 0.9388 | 35.24 / 0.7911 | 36.42 / 0.9253 |
Gaussian_Blur (MS & GT Blur) | 37.23 / 0.9552 | 30.34 / 0.9416 | 35.28 / 0.7650 | 42.10 / 0.8372 |
Pan_noise (PAN Noise) | 37.26 / 0.9535 | 30.19 / 0.9362 | 35.24 / 0.7735 | 43.67 / 0.9420 |
Removing the virtual equivariance loss (no_virtual) while keeping RST/PMA drops zero-shot PSNR by -2.07 dB on WV2, -1.60 dB on QB (HQNR drops to 0.8322), and -3.23 dB on GF2. Restricting SCT to fixed 8-band views (SCT Fixed 8) or discrete selection (SCT Selection) collapses GF2 PSNR by -7.76 dB and -7.58 dB and drops QB HQNR to 0.7619–0.7911. Adding non-equivariant spatial blur (Gaussian_Blur) or PAN noise (Pan_noise) violates task equivalence, lowering WV2/GF2 PSNR by 0.33–1.90 dB and collapsing QB HQNR to 0.7650 / 0.7735 (: 0.172 / 0.196 vs. 0.046).
3. Contribution of Reversible Mapping Beyond Simpler Mapping & Normalization (Q3)
As detailed in revised Section 3.3 and Appendix A, bridging bands precludes a global bijection on by rank-nullity. RST uses Helmert bases () to isolate common intensity from zero-mean chroma, combining orthogonalized embedding () with full Cholesky whitening to guarantee an exact analytical left inverse () and preserve task equivalence ( when ). We ablate RST_no_diagonal (per-channel variance scaling without off-diagonal Cholesky covariance ) and RST_no_stats (geometry alignment only, without statistical whitening/recoloring):
| Variant (ABFNet, WV3 Source) | WV3 PSNR / HQNR | WV2 PSNR / HQNR | QB PSNR / HQNR | GF2 PSNR / HQNR |
|---|---|---|---|---|
| EquiPan (Full RST) | 37.32 / 0.9528 | 31.85 / 0.9341 | 35.10 / 0.9121 | 44.00 / 0.9212 |
RST_no_diagonal (Per-channel std) | 37.37 / 0.9533 | 30.11 / 0.9371 | 35.14 / 0.8038 | 36.55 / 0.9297 |
RST_no_stats (Geometry only) | 37.41 / 0.9514 | 30.22 / 0.9287 | 34.76 / 0.7896 | 32.00 / 0.8894 |
Simplifying full Cholesky whitening to per-channel scaling (RST_no_diagonal) drops GF2 PSNR by -7.45 dB (to 36.55 dB), WV2 PSNR by -1.74 dB, and QB HQNR to 0.8038. Removing statistical alignment (RST_no_stats) collapses GF2 PSNR by -12.00 dB (to 32.00 dB).
4. Deployment Difference Between RST and Distribution-Alignment Methods (Q4)
As clarified in revised Section 2 and Section 3.3, distribution-alignment methods (e.g., UniPAN) apply 1D marginal CDF warping independently per input band while predicting native-space residuals, requiring fixed band counts and distorting inter-band ratios. In contrast, RST is a closed-loop forward-and-inverse subspace transport: target statistics estimated on stationary zero-mean chroma residuals not only whiten input chroma before the backbone, but also analytically recolor and project the predicted latent residual back to the native -band geometry via (Eq. 20), enabling lossless zero-shot transfer across arbitrary channel counts (8-to-4 and 4-to-8).
We sincerely thank Reviewer 8mng for the constructive comments. We have revised the manuscript (highlighted in blue) to incorporate your suggestions. Below we address all concerns and questions point by point.
1. Comparison with Closely Related Methods Under Comparable Conditions (W1 & Q1)