Pith. sign in

REVIEW 4 major objections 5 minor 62 references

DocShaDiffusion: Diffusion Model in Latent Space for Document Image Shadow Removal

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that latent-space diffusion with shadow-mask guidance removes document shadows, including colored ones, better than existing methods.

desk verdict The forward SDE contradicts the method's own pipeline, so this needs major revision despite a potentially useful dataset. read the letter →

arxiv 2507.01422 v1 pith:QA4GRO6I submitted 2025-07-02 cs.CV cs.AI

Wenjie Liu , Bingshu Wang , Ze Wang , C.L. Philip Chen This is my paper · ORCID
classification cs.CVcs.AI
keywords documentimageshadowremovallatentdiffusionmodelsoft-maskcolorshadowssyntheticdatasetmask-guidedshadow-robustperceptualloss
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper aims to make document image shadow removal work for colored shadows, not just the constant-color shadows that earlier methods target. It proposes DocShaDiffusion, a diffusion model that operates in latent space and uses a predicted shadow soft-mask to concentrate noise and denoising on shadowed regions, together with a shadow-robust perceptual loss. The authors also build SDCSRD, a 17,624-image synthetic dataset designed to mimic real color-shadow distributions, and report that their method outperforms prior state-of-the-art methods on the Adobe, FSDSRD, and SDCSRD benchmarks, with particularly large gains on their own dataset. The practical goal is a usable document shadow remover plus a training set that other methods can adopt.

What carries the argument

The load-bearing object is the shadow soft-mask $I^m$ produced by SSGM (Algorithm 1), a normalized map with values in $[0,1]$, set to 0 outside shadows and near 1 in the darkest shadow interior, obtained by sorting filtered grayscale pixels and thresholding with the mean of the darkest fraction. Inserted into the forward SDE $dx = \theta_t(\mu - x)\,dt - \sigma_t(1 - I^m)\,dw$, the mask is intended to concentrate random perturbation and denoising effort on shadow regions while leaving non-shadow areas largely untouched. The same mask guides a latent-space denoising network built from NAFBlocks with time embeddings, and the loss adds a VGG-19-based perceptual term that is designed to be less sensitive to shadow regions.

What would settle it

Take any SDCSRD shadow image for which the ground truth is known and measure the color of a shadowed patch before and after removing the shadow. If the synthetic shadow only scales RGB values by a scalar factor, the hue angle in the shadowed patch equals the hue in the ground-truth patch; a dataset that claims color shadows should show a shift in hue or saturation. A second check compares SSGM masks with manually annotated shadow boundaries on images with colored shadows on colored paper; low IoU on such images would mean the mask guidance fails exactly in the regime the paper targets.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that document shadow removal can be cast as a mask-guided latent diffusion problem. The shadow soft-mask generation module (SSGM) derives a soft mask from a single shadow image through grayscale conversion, dilation, median filtering, and luminance sorting, while the shadow mask-aware guided diffusion module (SMGDM) uses that mask to control the noise term in an IR-SDE forward process and to supervise the reverse denoising. Training couples the diffusion objective with a shadow-robust perceptual feature loss computed from VGG-19 features. According to the reported experiments, the resulting system outperforms existing heuristic and neural document shadow removal methods on Adobe, FSDSRD, and SDCSRD, with the largest margins on the synthetic color-shadow dataset it introduces.

Load-bearing premise

The load-bearing premise is that synthetic shadows generated by Eq. 1 with a scalar $C_k$ adequately represent real colored shadows, and that the luminance-sorting heuristic in SSGM can locate those shadows. If $C_k$ is truly scalar, the synthetic images are darker versions of the same hue, so the dataset contains no genuine color shadows and the paper's core motivation is not exercised.

Editorial extensions

If this is right

  • Document shadow removal can be formulated as a mask-guided latent diffusion problem, so shadow regions receive targeted noise and denoising while background detail is preserved.
  • The synthetic color-shadow dataset SDCSRD, generated with random shadow weight and shadow color parameters, gives other methods a large-scale training resource that existing synthetic datasets such as FSDSRD do not provide.
  • Because only shadow regions need to be denoised, the approach is expected to train faster and to avoid unnecessary changes to non-shadowed text and background.
  • The reported numbers on Adobe, FSDSRD, and SDCSRD indicate the model transfers to real document images without retraining on those sets, although on the RDD dataset it does not surpass CBENet, which was trained on RDD.
  • The ablation results suggest that the NAFBlock encoder-decoder, the mask-aware diffusion guidance, and the perceptual loss each contribute to the final performance, with the architecture change giving the largest single jump in PSNR and LPIPS.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If Eq. 1 is read literally with scalar $C_k$, the synthetic shadows are only darker versions of the same hue, so the dataset would not actually contain chromatic color shadows; the color-shadow story requires $C_k$ to be a color vector or channel-wise blending, which the text does not state.
  • The very large performance gap on SDCSRD compared with the more modest gains on RDD suggests that much of the reported advantage comes from matching the synthetic degradation model; a decisive test would be evaluation on real captured colored shadows on colored paper.
  • The same SSGM-plus-SMGDM recipe could plausibly transfer to other localized document degradations such as stains, watermarks, or demoiréing, since the mask mechanism is agnostic to the specific degradation type.
  • A direct reading of Eq. 5 gives noise coefficient $1-I^m$, which suppresses noise where the mask is 1, the opposite of the text's claim that noise is amplified in shadow regions; reconciling this sign would clarify what the mask is actually doing.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes DocShaDiffusion, a latent-space diffusion model for document image shadow removal. The method consists of a shadow soft-mask generation module (SSGM) that produces a soft mask from the luminance ordering of the input, and a shadow mask-aware guided diffusion module (SMGDM) built on the IR-SDE framework. The authors also introduce a synthetic dataset, SDCSRD, generated by a linear blending formula, and a shadow-robust perceptual feature loss based on VGG-19 features. Experiments are reported on Adobe, FSDSRD, SDCSRD, and RDD, with the central claim that DocShaDiffusion outperforms the state of the art, especially on color shadows. The paper states that code and dataset will be released.

Significance. If correct, the paper would make two contributions: a diffusion-based architecture tailored to document shadow removal, and a synthetic color-shadow dataset that could serve as training data for other methods. The ablation study shows a large positive effect of the proposed components (PSNR increases from 23.20 to 32.54 when all components are included), and the reported numbers on Adobe and FSDSRD are competitive with existing methods. However, the significance assessment is weakened by the absence of code/data, the lack of statistical validation, and, more importantly, by internal inconsistencies in the mathematical formulation of the core diffusion module and in the color-shadow dataset construction. The claimed SOTA result on the authors' own synthetic dataset is also not established under a fair comparison protocol.

major comments (4)
  1. [Section IV.B, Eqs. (5)-(6)] The forward SDE in Eq. (5) is internally inconsistent with the stated restoration goal. The text defines x as the shadowed image and mu as the shadow-free target. Under that assignment, the drift theta_t(mu - x) pushes the state toward the clean image, so X_T at the end of the forward process is approximately clean-plus-noise. The reverse SDE in Eq. (6) is the time reversal of that map and would send clean-plus-noise back toward shadowed images, not the other way around. The pipeline described in Section IV.A and Fig. 4 instead constructs X_T by adding mask-modulated noise to the shadowed input and then denoises to obtain a shadow-free image, which requires the opposite convention, namely mu equal to the shadowed observation, as in the cited Refusion work. Independently, the noise coefficient in Eq. (5), -sigma_t(1 - I_m)dw, vanishes where I_m = 1, i.e., in the darkest shadow regions, whereas the text in Section IV.B and the contribution bullet in the Introduction state that noise is added 'specially to shadow regions.' The two statements cannot both be true. This issue is load-bearing: as written, the core SMGDM module does not implement the described restoration process, and without released code or data the experimental tables cannot resolve which convention was actually used.
  2. [Section III.A, Eq. (1)] The dataset construction formula in Eq. (1) cannot generate color shadows as claimed. The text states that both a and C_k are scalar values, and I_sf, I_m are images in RGB color space. A scalar C_k multiplied by a scalar-valued mask (after the stated histogram equalization) scales all three color channels identically, producing grayscale shadow intensity shifts but no chromatic shift. Thus SDCSRD, as specified, is a gray-shadow dataset, not a color-shadow dataset, directly contradicting the paper's central motivation that existing methods 'ignore color shadows' and that SDCSRD 'simulates the distribution of realistic color shadows.' To produce colored shadows, C_k would need to be a three-dimensional vector (or otherwise channel-dependent). The authors should either correct Eq. (1) to include per-channel color factors or revise the claim that the dataset addresses color shadows.
  3. [Section V.B, Tables II and IV] The claim of outperforming the state of the art is not supported by the evaluation on SDCSRD. In Table II, DocShaDiffusion is trained on SDCSRD, while all baseline methods are evaluated with their original weights or provided results, as stated in Section V.B ('The results of other models are provided by MS-GAN, DCShadow-Net, CBENet'). The very large margins on SDCSRD (e.g., PSNR 41.66 vs. CBENet's 24.58) therefore largely reflect a training-set advantage, not model superiority. The averaging in Table IV mixes heterogeneous conditions: CBENet is trained on RDD while the proposed method is trained on SDCSRD. A valid SOTA claim requires a fair protocol, e.g., retraining all baselines on SDCSRD or testing all methods under the same cross-dataset conditions, or at minimum clearly labeling the tables as 'trained on SDCSRD' vs. 'not trained on SDCSRD.' Without this, the central 'superiority over state-of-the-art' assertion is not established.
  4. [Section V.A and V.C] The paper reports no error bars, repeated runs, or significance tests for any of the quantitative claims, including the statements in Section V.C that NAFBlock, SMGDM, and L_fea 'play important role' and 'prove beneficial.' Additionally, the feature-loss weights w_i in Eq. (8) are never specified, so the loss in Eq. (9) is incompletely defined. Given the single-run nature of the tables and the unspecified weights, the 'significantly outperforms' language in Section V.B and the conclusion is not backed by statistical evidence.
minor comments (5)
  1. [Throughout] There are numerous typographical errors and inconsistencies: 'datset' for dataset, 'tansformer' for transformer, 'P SN R' spacing artifacts in Tables I-III, and 'the k-th slice (k = 0, 1, ... , 4)' in Section IV.C, where k should range over 0..4 consistently.
  2. [Section IV.C, Eq. (7)] The notation is unclear: the loss uses ∥e_t − ϵ∥, where e_t is described as the predicted noise map, but e_t is not defined in the text; presumably it should be ϵθ, the neural network's noise prediction. Please clarify.
  3. [Fig. 3 and Section III.B] The dataset 'HS' is referenced in the caption of Fig. 3 and in Section III.B, but it is never introduced or defined in the paper. Please identify which real dataset is meant.
  4. [Tables I and II] The venue/year for 'Our Method' is listed as 'CVPR/2024,' which is not appropriate for a submitted manuscript and does not match the paper's arXiv date of July 2025. This should be corrected to avoid confusing the reader.
  5. [Section IV.B] The sentence 'The experimental results indicate that executing the denoising process in pixel space would necessitate excessive computational resources' reports a result that is never shown; either provide the supporting experiment or remove the unsupported claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the paper's central claim is empirical and externally benchmarked; the noted SDE and mask issues are correctness concerns, not input-output equivalence.

full rationale

DocShaDiffusion's derivation chain does not reduce to its inputs. The SSGM shadow soft-mask is a heuristic preprocessing step (Algorithm 1 and Eqs. 2-4), not a fitted quantity renamed as a prediction. The SMGDM inherits the IR-SDE forward and reverse equations from external prior work (Refusion, reference [54]), and the training objective is a standard diffusion noise-prediction loss (Eq. 7) combined with a VGG-based perceptual loss (Eq. 9); no parameter is fitted to the target and then reported as a prediction. The main empirical claim is tested against external benchmarks (Adobe, FSDSRD, and RDD) as well as the authors' own SDCSRD, so the state-of-the-art comparison is not statistically forced by construction. The paper's self-citations ([5], [17], [18], [33]) refer to prior heuristic baselines and are not load-bearing for the diffusion design. The internal inconsistencies noted by the reviewer (Eq. 1 using a scalar Ck to model color shadows, and the apparent forward/reverse semantics of Eqs. 5-6 together with the (1-I_m) mask weighting) are correctness or clarity risks rather than circularity, because the text never identifies a predicted quantity with a fitted input. The absence of released code and data is a reproducibility limitation, not a circular step. Therefore no circular step can be quoted or exhibited, and the appropriate finding is no significant circularity.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The central claim rests on borrowed diffusion math, a luminance-based shadow detector, a synthetic-data generation model whose scalar C_k conflicts with color-shadow claims, and an under-specified feature loss. No new theoretical entities such as particles or forces are introduced; SDCSRD is an empirical dataset, not a postulated entity. Independent evidence for the synthetic shadow realism and for the shadow-invariance of deep VGG features is not supplied.

free parameters (5)
  • lambda (loss weight) = 0.5
    Weight balancing the diffusion objective and the feature loss, chosen by hand in Section V.A.
  • diffusion noise level = 50
    Fixed noise level for the latent diffusion process, chosen by hand in Section V.A.
  • diffusion denoising steps = 100
    Number of reverse denoising steps, chosen by hand in Section V.A.
  • SSGM low-intensity fraction a = not specified
    Eq. 2 selects a fraction of the darkest pixels for the shadow mask, but the value of a is never reported.
  • VGG feature weights w_i = not specified
    Eq. 8 defines a weighted sum of VGG-19 slice features, but the weights are never stated, derived, or cited, making the feature loss under-specified.
assumptions (5)
  • standard math IR-SDE forward and reverse models with mask-dependent noise are a valid generative model for shadow removal.
    Borrowed from Luo et al. [54] and treated as given in Eqs. 5 and 6 without derivation or validation in this document setting.
  • domain assumption Shadows can be localized by luminance sorting after grayscale conversion, dilation, and median filtering.
    Algorithm 1 in Section IV.A assumes this heuristic works for colored shadows and textured document backgrounds, but no validation is provided.
  • domain assumption Linear alpha-blending in Eq. 1 with a scalar shadow color simulates realistic color-shadow distributions.
    Section III.A states C_k is scalar, which contradicts the color-shadow claim; the resemblance to real data is asserted only through histograms in Fig. 3.
  • domain assumption Deep VGG-19 features are progressively less affected by shadows and can be combined into a shadow-robust perceptual loss.
    Section IV.C and Fig. 5 make this observation, but the weights w_i and the invariance claim are not independently tested.
  • domain assumption Latent-space diffusion in a NAFBlock-based autoencoder preserves the document details needed for reconstruction.
    Section IV.B justifies the latent-space choice by computational cost, not by a completeness or reconstruction guarantee.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DocShaDiffusion: Diffusion Model in Latent Space for Document Image Shadow Removal." pith.science (2026). https://pith.science/paper/QA4GRO6I

@misc{pith2026250701422,
  author       = {Pith},
  title        = {Pith review of: DocShaDiffusion: Diffusion Model in Latent Space for Document Image Shadow Removal},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QA4GRO6I}},
  note         = {Machine review of arXiv:2507.01422}
}
read the original abstract

Document shadow removal is a crucial task in the field of document image enhancement. However, existing methods tend to remove shadows with constant color background and ignore color shadows. In this paper, we first design a diffusion model in latent space for document image shadow removal, called DocShaDiffusion. It translates shadow images from pixel space to latent space, enabling the model to more easily capture essential features. To address the issue of color shadows, we design a shadow soft-mask generation module (SSGM). It is able to produce accurate shadow mask and add noise into shadow regions specially. Guided by the shadow mask, a shadow mask-aware guided diffusion module (SMGDM) is proposed to remove shadows from document images by supervising the diffusion and denoising process. We also propose a shadow-robust perceptual feature loss to preserve details and structures in document images. Moreover, we develop a large-scale synthetic document color shadow removal dataset (SDCSRD). It simulates the distribution of realistic color shadows and provides powerful supports for the training of models. Experiments on three public datasets validate the proposed method's superiority over state-of-the-art. Our code and dataset will be publicly available.

Figures

Figures reproduced from arXiv: 2507.01422 by the authors.

Figure 1
Figure 1. An example of our proposed DocShaDiffusion for [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Examples of our dataset (SDCSRD). From up to down: [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Comparison of color distribution histograms between [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Illustration of the proposed DocShaDiffusion, in which the training process is dashed line and sampling process is solid [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Illustration of the proposed Lf ea. Is is shadow image. Isf is shadow-free image. Vi is the feature map in VGG-19. [60]. We observe that with an increase in the number of convolutional layers in the VGG-19 network, the extracted feature map, as shown in Fig.5, becomes …
Figure 6
Figure 6. Figure 6: Visual comparison between our method and the SOTA methods. (a)input images, (b)ISR-based [32], (c)3D-PC-based [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 56 canonical work pages

  1. [1]

    A statistical approach for multi-frame shadow movement detection and shadow removal for document capture,

    P. Mondal and A. Bal, “A statistical approach for multi-frame shadow movement detection and shadow removal for document capture,” in 30th European Signal Processing Conference . IEEE, 2022, pp. 508–512

    work page 2022

  2. [2]

    Lp-ioanet: Efficient high resolution document shadow removal,

    K. Georgiadis, M. K. Yucel, E. Skartados, V . Dimaridou, A. Drosou, A. Saa-Garriga, and B. Manganelli, “Lp-ioanet: Efficient high resolution document shadow removal,” in 2023 IEEE International Conference on Acoustics, Speech and Signal Processing . IEEE, 2023, pp. 1–5

    work page 2023

  3. [3]

    Ddoce: Deep document enhancement with multi-scale feature aggregation and pixel-wise adjustments,

    K. O. Bogdan, G. A. Megeto, R. Leal, G. Souza, A. C. Valente, and L. N. Kirsten, “Ddoce: Deep document enhancement with multi-scale feature aggregation and pixel-wise adjustments,” in International Conference on Document Analysis and Recognition . Springer, 2021, pp. 229–244

    work page 2021

  4. [4]

    A boundary-aware network for shadow removal,

    K. Niu, Y . Liu, E. Wu, and G. Xing, “A boundary-aware network for shadow removal,” IEEE Transactions on Multimedia , 2022

    work page 2022

  5. [5]

    Moving cast shadows segmentation using illumination invariant feature,

    B. Wang, Y . Zhao, and C. P. Chen, “Moving cast shadows segmentation using illumination invariant feature,” IEEE Transactions on Multimedia, vol. 22, no. 9, pp. 2221–2233, 2019

    work page 2019

  6. [6]

    Document rectification and illumination correction using a patch-based cnn,

    X. Li, B. Zhang, J. Liao, and P. V . Sander, “Document rectification and illumination correction using a patch-based cnn,” ACM Transactions on Graphics, vol. 38, no. 6, pp. 1–11, 2019

    work page 2019

  7. [7]

    Deepotsu: Document enhancement and bina- rization using iterative deep learning,

    S. He and L. Schomaker, “Deepotsu: Document enhancement and bina- rization using iterative deep learning,” in Pattern Recognition, vol. 91. Elsevier, 2019, pp. 379–390

    work page 2019

  8. [8]

    Binarization of degraded document images with global-local u-nets,

    X. Huang, L. Li, R. Liu, C. Xu, and M. Ye, “Binarization of degraded document images with global-local u-nets,” Optik, vol. 203, p. 164025, 2020

    work page 2020

Show all 62 references
  1. [9]

    Docstormer: Revitalizing multi-degraded colored document images to pristine pdf,

    C. Liu, J. Li, Y . Teng, C. Wang, N. Xu, J. Wu, and D. Tu, “Docstormer: Revitalizing multi-degraded colored document images to pristine pdf,” arXiv preprint arXiv:2310.17910 , 2023

    2023 arXiv

  2. [10]

    Light-weight document image cleanup using perceptual loss,

    S. Dey and P. Jawanpuria, “Light-weight document image cleanup using perceptual loss,” in Document Analysis and Recognition . Springer, 2021, pp. 238–253

    2021

  3. [11]

    Document image binarization using visibility detection and point cloud segmentation,

    J. Li, Y . Chen, and S. Liu, “Document image binarization using visibility detection and point cloud segmentation,” in Pattern Recognition and Computer Vision. Springer, 2021, pp. 92–104

    2021

  4. [12]

    Appearance en- hancement for camera-captured document images in the wild,

    J. Zhang, L. Liang, K. Ding, F. Guo, and L. Jin, “Appearance en- hancement for camera-captured document images in the wild,” IEEE Transactions on Artificial Intelligence , 2023

    2023

  5. [13]

    Shadow re- moval from document image based on background estimation employing selective median filter and black-top-hat transform,

    S. Imahayashi, M. Mukaida, S. Takeda, and N. Suetake, “Shadow re- moval from document image based on background estimation employing selective median filter and black-top-hat transform,” Optical Review, pp. 1–5, 2023

    2023

  6. [14]

    Udoc-gan: Unpaired document illumination correction with background light prior,

    Y . Wang, W. Zhou, Z. Lu, and H. Li, “Udoc-gan: Unpaired document illumination correction with background light prior,” in Proceedings of the 30th ACM International Conference on Multimedia, 2022, pp. 5074– 5082

    2022

  7. [15]

    Docde- shadower: Frequency-aware transformer for document shadow removal,

    S. Luo, R. Xu, X. Chen, Z. Li, C.-M. Pun, and S. Wang, “Docde- shadower: Frequency-aware transformer for document shadow removal,” arXiv preprint arXiv:2307.15318 , 2023

    2023 arXiv

  8. [16]

    Removing shadows from images of documents,

    S. Bako, S. Darabi, E. Shechtman, J. Wang, K. Sunkavalli, and P. Sen, “Removing shadows from images of documents,” in Asian Conference on Computer Vision , 2016, pp. 173–183

    2016

  9. [17]

    Joint water-filling algorithm with adaptive chroma adjustment for shadow removal from text document images,

    Z. Wang, B. Wang, J. Zheng, and C. Philip Chen, “Joint water-filling algorithm with adaptive chroma adjustment for shadow removal from text document images,” in IEEE International Conference on Systems, Man, and Cybernetics , 2022, pp. 2882–2887

    2022

  10. [18]

    Shadow removal of text document images using background estimation and adaptive text enhancement,

    W. Liu, B. Wang, J. Zheng, and W. Wang, “Shadow removal of text document images using background estimation and adaptive text enhancement,” in 2023 IEEE International Conference on Acoustics, Speech and Signal Processing . IEEE, 2023, pp. 1–5

    2023

  11. [19]

    Document enhancement using visibility detection,

    N. Kligler, S. Katz, and A. Tal, “Document enhancement using visibility detection,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 2374–2382

    2018

  12. [20]

    Water-filling: An efficient algorithm for digitized document shadow removal,

    S. Jung, M. A. Hasan, and C. Kim, “Water-filling: An efficient algorithm for digitized document shadow removal,” in Asian Conference on Computer Vision. Springer, 2018, pp. 398–414

    2018

  13. [21]

    A survey on deep learning based document image enhancement,

    Z. Anvari and V . Athitsos, “A survey on deep learning based document image enhancement,” arXiv preprint arXiv:2112.02719 , 2021

    2021 arXiv

  14. [22]

    Unsupervised learning for intrinsic image decomposition from a single image,

    Y . Liu, Y . Li, S. You, and F. Lu, “Unsupervised learning for intrinsic image decomposition from a single image,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2020, pp. 3248–3257

    2020

  15. [23]

    Shadow removal via shadow image decompo- sition,

    H. Le and D. Samaras, “Shadow removal via shadow image decompo- sition,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 8578–8587

    2019

  16. [24]

    Self- supervised shadow removal,

    F.-A. Vasluianu, A. Romero, L. Van Gool, and R. Timofte, “Self- supervised shadow removal,” arXiv preprint arXiv:2010.11619 , 2020

    2010 arXiv

  17. [25]

    Shadocnet: Learning spatial-aware tokens in transformer for document shadow removal,

    X. Chen, X. Cun, C.-M. Pun, and S. Wang, “Shadocnet: Learning spatial-aware tokens in transformer for document shadow removal,” in 2023 IEEE International Conference on Acoustics, Speech and Signal Processing. IEEE, 2023, pp. 1–5

    2023

  18. [26]

    Shadoc- former: A shadow-attentive threshold detector with cascaded fusion re- finer for document shadow removal,

    W. Chen, S. Luo, X. Chen, Z. Li, S. Wang, and C.-M. Pun, “Shadoc- former: A shadow-attentive threshold detector with cascaded fusion re- finer for document shadow removal,” arXiv preprint arXiv:2309.06670 , 2023

    2023 arXiv

  19. [27]

    High-resolution document shadow removal via a large-scale real-world dataset and a frequency- aware shadow erasing net,

    Z. Li, X. Chen, C.-M. Pun, and X. Cun, “High-resolution document shadow removal via a large-scale real-world dataset and a frequency- aware shadow erasing net,” in Proceedings of the IEEE/CVF Inter- national Conference on Computer Vision , October 2023, pp. 12 449– 12 458

    2023

  20. [28]

    A fast and efficient network for single image shadow detection,

    L. Jie and H. Zhang, “A fast and efficient network for single image shadow detection,” in IEEE International Conference on Acoustics, Speech and Signal Processing . IEEE, 2022, pp. 2634–2638

    2022

  21. [29]

    Bedsr-net: A deep shadow removal network from a single document image,

    Y .-H. Lin, W.-C. Chen, and Y .-Y . Chuang, “Bedsr-net: A deep shadow removal network from a single document image,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2020, pp. 12 905–12 914

    2020

  22. [30]

    Document image shadow removal guided by color-aware background,

    L. Zhang, Y . He, Q. Zhang, Z. Liu, X. Zhang, and C. Xiao, “Document image shadow removal guided by color-aware background,” in Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 1818–1827

    2023

  23. [31]

    Document shadow removal with foreground detection learning from fully synthetic images,

    Y . Matsuo, N. Akimoto, and Y . Aoki, “Document shadow removal with foreground detection learning from fully synthetic images,” in 2022 IEEE International Conference on Image Processing . IEEE, 2022, pp. 1656–1660

    2022

  24. [32]

    An iterative approach for shadow removal in document images,

    V . Shah and V . Gandhi, “An iterative approach for shadow removal in document images,” in 2018 IEEE International Conference on Acoustics, Speech and Signal Processing . IEEE, 2018, pp. 1892–1896. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 9

    2018

  25. [33]

    Local water-filling algorithm for shadow detection and removal of document images,

    B. Wang and C. P. Chen, “Local water-filling algorithm for shadow detection and removal of document images,” Sensors, vol. 20, no. 23, p. 6929, 2020

    2020

  26. [34]

    Video-based document image scanning using a mobile device,

    B. Jiang, S. Liu, S. Xia, X. Yu, M. Ding, X. Hou, and Y . Gao, “Video-based document image scanning using a mobile device,” in 2015 International Conference on Image and Vision Computing New Zealand. IEEE, 2015, pp. 1–6

    2015

  27. [35]

    Shadow removal of text document images by estimating local and global background colors,

    J.-R. Wang and Y .-Y . Chuang, “Shadow removal of text document images by estimating local and global background colors,” in 2020 IEEE International Conference on Acoustics, Speech and Signal Processing . IEEE, 2020, pp. 1534–1538

    2020

  28. [36]

    Docdiff: Document enhancement via residual diffusion models,

    Z. Yang, B. Liu, Y . Xxiong, L. Yi, G. Wu, X. Tang, Z. Liu, J. Zhou, and X. Zhang, “Docdiff: Document enhancement via residual diffusion models,” in Proceedings of the 31st ACM International Conference on Multimedia, 2023, pp. 2795–2806

    2023

  29. [37]

    Shadowdiffusion: When degradation prior meets diffusion model for shadow removal,

    L. Guo, C. Wang, W. Yang, S. Huang, Y . Wang, H. Pfister, and B. Wen, “Shadowdiffusion: When degradation prior meets diffusion model for shadow removal,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 14 049–14 058

    2023

  30. [38]

    Cross: Diffusion model makes controllable, robust and secure image steganography,

    J. Yu, X. Zhang, Y . Xu, and J. Zhang, “Cross: Diffusion model makes controllable, robust and secure image steganography,” arXiv preprint arXiv:2305.16936, 2023

    2023 arXiv

  31. [39]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” in Advances in neural information processing systems, vol. 33, 2020, pp. 6840–6851

    2020

  32. [40]

    Improved denoising diffusion probabilis- tic models,

    A. Q. Nichol and P. Dhariwal, “Improved denoising diffusion probabilis- tic models,” in International Conference on Machine Learning. PMLR, 2021, pp. 8162–8171

    2021

  33. [41]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” arXiv preprint arXiv:2010.02502 , 2020

    2010 arXiv

  34. [42]

    Denoising diffusion restoration models,

    B. Kawar, M. Elad, S. Ermon, and J. Song, “Denoising diffusion restoration models,” in Advances in Neural Information Processing Systems, vol. 35, 2022, pp. 23 593–23 606

    2022

  35. [43]

    Image super-resolution via iterative refinement,

    C. Saharia, J. Ho, W. Chan, T. Salimans, D. J. Fleet, and M. Norouzi, “Image super-resolution via iterative refinement,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 45, no. 4, pp. 4713– 4726, 2022

    2022

  36. [44]

    Repaint: Inpainting using denoising diffusion probabilistic models,

    A. Lugmayr, M. Danelljan, A. Romero, F. Yu, R. Timofte, and L. Van Gool, “Repaint: Inpainting using denoising diffusion probabilistic models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 11 461–11 471

    2022

  37. [45]

    Palette: Image-to-image diffusion models,

    C. Saharia, W. Chan, H. Chang, C. Lee, J. Ho, T. Salimans, D. Fleet, and M. Norouzi, “Palette: Image-to-image diffusion models,” in ACM SIGGRAPH 2022 Conference Proceedings , 2022, pp. 1–10

    2022

  38. [46]

    Coldbin: Cold diffusion for document image binarization,

    S. Saifullah, S. Agne, A. Dengel, and S. Ahmed, “Coldbin: Cold diffusion for document image binarization,” in International Conference on Document Analysis and Recognition . Springer, 2023, pp. 207–226

    2023

  39. [47]

    Raindiffusion: When unsupervised learning meets diffusion models for real-world image deraining,

    M. Wei, Y . Shen, Y . Wang, H. Xie, and F. L. Wang, “Raindiffusion: When unsupervised learning meets diffusion models for real-world image deraining,” arXiv preprint arXiv:2301.09430 , 2023

    2023 arXiv

  40. [48]

    Frequency compensated diffusion model for real-scene dehazing,

    J. Wang, S. Wu, K. Xu, and Z. Yuan, “Frequency compensated diffusion model for real-scene dehazing,” arXiv preprint arXiv:2308.10510, 2023

    2023 arXiv

  41. [49]

    Learning from synthetic shadows for shadow detection and removal,

    N. Inoue and T. Yamasaki, “Learning from synthetic shadows for shadow detection and removal,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 31, no. 11, pp. 4187–4197, 2020

    2020

  42. [50]

    Revisiting shadow detection: A new benchmark dataset for complex world,

    X. Hu, T. Wang, C.-W. Fu, Y . Jiang, Q. Wang, and P.-A. Heng, “Revisiting shadow detection: A new benchmark dataset for complex world,” IEEE Transactions on Image Processing , vol. 30, pp. 1925– 1934, 2021

    1925

  43. [51]

    Instance shadow detection with a single-stage detector,

    T. Wang, X. Hu, P.-A. Heng, and C.-W. Fu, “Instance shadow detection with a single-stage detector,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 3, pp. 3259–3273, 2022

    2022

  44. [52]

    Distraction-aware shadow detection,

    Q. Zheng, X. Qiao, Y . Cao, and R. W. Lau, “Distraction-aware shadow detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2019, pp. 5167–5176

    2019

  45. [53]

    A multi-task mean teacher for semi-supervised shadow detection,

    Z. Chen, L. Zhu, L. Wan, S. Wang, W. Feng, and P.-A. Heng, “A multi-task mean teacher for semi-supervised shadow detection,” in Proceedings of the IEEE/CVF Conference on computer vision and pattern recognition, 2020, pp. 5611–5620

    2020

  46. [54]

    Refusion: Enabling large-size realistic image restoration with latent- space diffusion models,

    Z. Luo, F. K. Gustafsson, Z. Zhao, J. Sj ¨olund, and T. B. Sch ¨on, “Refusion: Enabling large-size realistic image restoration with latent- space diffusion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 1680–1691

    2023

  47. [55]

    High- resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High- resolution image synthesis with latent diffusion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion, 2022, pp. 10 684–10 695

    2022

  48. [56]

    Simple baselines for image restoration,

    L. Chen, X. Chu, X. Zhang, and J. Sun, “Simple baselines for image restoration,” in European Conference on Computer Vision . Springer, 2022, pp. 17–33

    2022

  49. [57]

    An effective background estimation method for shadows removal of document images,

    B. Wang and C. P. Chen, “An effective background estimation method for shadows removal of document images,” in 2019 IEEE International Conference on Image Processing . IEEE, 2019, pp. 3611–3615

    2019

  50. [58]

    Mask-shadowgan: Learn- ing to remove shadows from unpaired data,

    X. Hu, Y . Jiang, C.-W. Fu, and P.-A. Heng, “Mask-shadowgan: Learn- ing to remove shadows from unpaired data,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 2472– 2481

    2019

  51. [59]

    Dc-shadownet: Single-image hard and soft shadow removal using unsupervised domain-classifier guided network,

    Y . Jin, A. Sharma, and R. T. Tan, “Dc-shadownet: Single-image hard and soft shadow removal using unsupervised domain-classifier guided network,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 5027–5036

    2021

  52. [60]

    Bijective mapping network for shadow removal,

    Y . Zhu, J. Huang, X. Fu, F. Zhao, Q. Sun, and Z.-J. Zha, “Bijective mapping network for shadow removal,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 5627–5636

    2022

  53. [61]

    Unsupervised shadow removal using target consis- tency generative adversarial network,

    C. Tan and X. Feng, “Unsupervised shadow removal using target consis- tency generative adversarial network,” arXiv preprint arXiv:2010.01291, 2020. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 10 Wenjie Liu received her M.S. degree in software engineering from Nor...

    2010 arXiv

  54. [2013]

    He is also an member of Chinese Association of Automation (CAA), China Computer Federation (CCF), Chinese Association for Artificial Intelligence(CAAI)

    Now he is an associate professor in School of Software, Northwestern Polytechnical Univer- sity. He is also an member of Chinese Association of Automation (CAA), China Computer Federation (CCF), Chinese Association for Artificial Intelligence(CAAI). His current research intere...

    2024

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.