- Rating
- 3(+1)
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 3
Read review
Paper summary
This paper proposes Learning to Recorrupt (L2R), a self supervised image denoising framework for settings where the exact noise distribution is unknown. The main idea is to replace the fixed recorruption process used in previous recorruption based methods with a learnable recorruptor. The denoiser and recorruptor are jointly optimized through a min max objective. The learned recorruptor is designed to capture auxiliary noise statistics that make the self supervised objective approximate the corresponding supervised denoising objective. The paper also establishes connections between L2R, GR2R, and UNSURE, and introduces different recorruptor parameterizations for independent, spatially correlated, and signal dependent noise. Experiments on synthetic noise, SIDD, and Cryo EM data evaluate the proposed framework under a range of noise conditions.
Critical strengths
- The paper provides a principled way to learn the recorruption process instead of assuming that the exact noise distribution is known in advance. The min max formulation gives a clear connection between the learned recorruption and the supervised denoising objective.
- The experimental study is relatively thorough. The paper evaluates different noise structures, analyzes the recorruptor design and optimization strategy, and includes additional studies on training stability and learned noise statistics.
Critical weaknesses
-
Limited methodological novelty relative to GR2R and UNSURE. The proposed method is closely connected to existing recorruption based objectives. The paper itself describes L2R as a learned extension of GR2R and a learned generalization of UNSURE. The main change is to replace a fixed or analytically specified recorruption mechanism with a learnable neural model that is optimized jointly with the denoiser. This is useful, but I am not fully convinced that the resulting methodological advance is large enough for ICLR.
-
The claim of noise distribution agnostic denoising appears stronger than the actual assumptions of the method. Although L2R does not require the exact noise distribution or its parameters, the recorruptor architecture still needs to reflect the broad structure of the corruption process. Different designs are used for independent, spatially correlated, and signal dependent noise. The appendix also states that the recorruptor architecture acts as a user specified structural prior. Therefore, the method is not fully agnostic to the noise structure, and the current wording may overstate its generality.
Questions for authors
- Since L2R is presented as a learned extension of GR2R and is also closely related to UNSURE, could the authors clarify what new learning principle is introduced beyond replacing the fixed recorruption mechanism with a learned one?
- The method uses different recorruptor structures for independent, correlated, and signal dependent noise. How should a practitioner select the appropriate recorruptor when the noise structure itself is unknown?
- Could the authors clarify the practical failure cases when the chosen recorruptor class does not contain a compatible noise transformation? How sensitive is the method to such model mismatch?
- Could the authors provide comparisons with more recent self supervised denoising methods to better position L2R with respect to the current state of the field?
7 comments
Response to Reviewer 3p15 (2/2)
W2 "The claim of noise distribution agnostic denoising appears stronger than the actual assumptions of the method. Although L2R does not require the exact noise distribution or its parameters, the recorruptor architecture still needs to reflect the broad structure of the corruption process. Different designs are used for independent, spatially correlated, and signal dependent noise. The appendix also states that the recorruptor architecture acts as a user specified structural prior. Therefore, the method is not fully agnostic to the noise structure, and the current wording may overstate its generality."
We agree, and we corrected this claim in the revised paper: the abstract now states that L2R does not require knowledge of the noise distribution or its parameters, only of its broad structure. This structure is typically known from the acquisition pipeline, e.g., whether preprocessing is applied and the sensor type.
Q2 "The method uses different recorruptor structures for independent, correlated, and signal dependent noise. How should a practitioner select the appropriate recorruptor when the noise structure itself is unknown?"
In practice, the structure is typically known from the acquisition: whether an ISP or other preprocessing is applied indicates spatial correlation, and the sensor and acquisition regime (e.g., photon-limited CMOS imaging, as in SIDD, or SPAD sensors) indicates signal dependence. Otherwise, it can be diagnosed from noisy data alone. Spatial autocorrelation of the residual indicates the need for the spatial kernel. Dependence of the residual variance on intensity indicates the need for . We added this guidance to App. A.2.2.
Q3.1 "Could the authors clarify the practical failure cases when the chosen recorruptor class does not contain a compatible noise transformation?"
Table 7 (appendix) quantifies this; the bold rows are the recorruptor class matching each noise structure:
| Noise | Kernel | PSNR | SSIM | |
|---|---|---|---|---|
| Log-gamma | 1×1 | no | 30.77 | 0.871 |
| 3×3 | no | 29.93 | 0.837 | |
| 1×1 | yes | 26.98 | 0.721 | |
| 3×3 | yes | 24.02 | 0.657 | |
| Correlated | 1×1 | no | 24.21 | 0.660 |
| 3×3 | no | 29.85 | 0.863 | |
| 1×1 | yes | 24.02 | 0.657 | |
| 3×3 | yes | 27.05 | 0.754 | |
| Poisson–Gaussian | 1×1 | no | 24.99 | 0.628 |
| 3×3 | no | 23.77 | 0.581 | |
| 1×1 | yes | 27.80 | 0.791 | |
| 3×3 | yes | 26.03 | 0.713 |
The main failure mode is a recorruptor class that lacks the structure of the noise, e.g., no spatial kernel for correlated noise (−5.64 dB) or no signal-dependent scale for Poisson–Gaussian noise (−2.81 dB).
Q3.2 "How sensitive is the method to such model mismatch?"
An unnecessary spatial kernel costs only 0.84 dB (Table 7). Within the correct class, L2R is robust to , to the learning-rate ratio, and to the size of (Tables 6, 8, 9).
Q4 "Could the authors provide comparisons with more recent self supervised denoising methods to better position L2R with respect to the current state of the field?"
We thank the reviewer for this suggestion. Our comparison already includes the latest and most competitive self-supervised methods that are relevant to our setting. These include UNSURE (ICLR 2025) and GR2R (CVPR 2025), the current state of the art among SURE- and recorruption-based methods. In particular, within our RAW SIDD protocol, Blind2Sound (ICCV 2025) is the most recent and strongest competitor, and it is included in Tables 1, 2, and 3.
We hope our response and the revised paper (changes in blue) have addressed your concerns, in particular the new comparison with learning h directly within the GR2R loss (App. A.3.4). Please let us know if you have any questions regarding our response.
Response to Reviewer 3p15 (1/2)
We thank the reviewer for this important question. We would like to clarify why making the recorruption learnable is not a simple replacement:
To support the first point, we compared three ways of learning h in the log-gamma ℓ=1 setting of Table 4 (BSDS500), with the same denoiser, recorruptor class, data, and training budget (now added to App. A.3.4):
Both naive objectives fail, as expected from the relation between the two losses (Eq. 35, App. A.1.1). Since y2=y−h(w′)/τ and h is zero-mean and independent of y,
E∥f(y1)−y2∥22=E[∥f(y1)−y∥22+τ2f(y1)⊤h(w′)]+τ21E∥h(w′)∥22,
where the first term is the L2R loss, i.e., the GR2R loss equals the L2R loss plus an amplitude term. Maximizing the GR2R loss over h is dominated by this term: the learned noise scale keeps growing during training, from 0.1 to 3.48 (about 35 times the noise level). Minimizing it instead drives h toward a degenerate distribution, and the result falls below the noisy input (20.00 dB). L2R removes exactly this amplitude term, so that h learns the noise statistics rather than its scale.
Beyond this new principle, L2R learns the recorruption from noisy data alone, enforces a richer family of finite-τ constraints than UNSURE's single divergence (recovering UNSURE as the linear case, App. A.1.2), needs no divergence estimation (one denoiser evaluation per sample, Algorithm 1), and covers independent, correlated, and signal-dependent noise with one loss.
Empirically, L2R improves over UNSURE by up to 5.7 dB on heavy-tailed noise (Table 1). L2R also improves by +3.15 dB over a finite-τ UNSURE (learned Gaussian scale, Table 4). It is also more efficient: on the Cryo-EM data, L2R trains 1.55× faster than PG-UNSURE under matched conditions.