Full search page →

Bilateral grids and color-table mathematics

Optional · supplementary note 12 of 14 for DLCV 09C · Low-Light Image Enhancement. Nothing in the main text depends on it; citation numbers refer to the reference list at the bottom of this page.

Navigate: this page and this track: Navigate: this page and this track

On this page

Supplementary notes · 14

  1. CLAHE: equalizing locally without amplifying noise
  2. More deep LLIE methods
  3. Counting resolution, arithmetic and memory
  4. A bilateral-grid experiment
  5. Implementing and checking trilinear LUTs
  6. Fitting and slicing a bilateral grid in detail
  7. Learnable LUT variants in detail
  8. Structural re-parameterization step by step
  9. Understanding no-reference low-light metrics
  10. Efficient network building blocks
  11. Banding, halos and temporal stability
  12. Bilateral grids and color-table mathematics
  13. From retouching to low light
  14. A lightness-order exercise

Paper guides · 13

  1. KinD: repair reflectance while controlling illumination
  2. Retinex-Net: decompose, adjust and denoise
  3. EnlightenGAN: unpaired appearance learning with content constraints
  4. Zero-DCE and Zero-DCE++: learn curves without reference images
  5. RUAS: search the priors inside an unrolled enhancement model
  6. SCI: train with calibration, deploy one illumination block
  7. Retinexformer: illumination guides restoration
  8. IAT: local affine maps and a global camera-style transform
  9. Multinex: fuse analytic priors into a Retinex residual
  10. CPGA-Net: channel priors, gamma estimation and learned fusion
  11. CPGA-Net+: attention, BDSF and stronger variants
  12. HDRNet: deep bilateral learning for real-time image enhancement
  13. Image-adaptive 3D LUTs (Zeng et al., TPAMI 2022)

Deep Learning for Computer Vision

  1. 02 · Deep Image Classification
View the whole course tree →

Plain version. The last lineage comes from photo editing rather than from low-light research. Its goal was to reproduce a photographer’s retouching, or a camera’s processing pipeline, in real time on full-resolution photos. Its two tools, a grid of small color recipes and a color lookup table, turned out to be among the cheapest ways to apply any smooth color transform, and they were later used for underexposed and low-light images as well.

Where it comes from. The methods in this section were trained and evaluated on photo-retouching data, as their own papers state. HDRNet [28] learns from MIT-Adobe FiveK [71] (5,000 raw photos, each retouched by five experts) and from the outputs of photographic operators, among them the HDR+ burst-photography pipeline [93] and Photoshop actions. Image-adaptive 3D LUTs [29] are trained on FiveK (expert C as the target) and on the HDR+ burst-photography dataset, whose finished photos come from a camera’s HDR pipeline. AdaInt [43] and SepLUT [44] use FiveK and a large portrait-retouching dataset. The Li et al. survey leaves such general photo-enhancement methods out, as not designed for diverse low-light conditions, and argues that FiveK’s references do not suit low-light enhancement [15].

Bilateral grids and guided upsampling

Plain version. Imagine cutting the image into large tiles, and inside each tile, also sorting the pixels into a few brightness levels. For each (tile, brightness level) box, store one small color recipe. A pixel finds its box by where it is and how bright it is, so two neighboring pixels on opposite sides of an edge (one dark, one bright) use different recipes, and the edge stays sharp even though the tiles are big.

Precise version. A bilateral grid [94] is a 3D array indexed by two coarse spatial coordinates and one coarse intensity coordinate. Chen, Paris and Durand introduced it to accelerate edge-aware filters such as the bilateral filter [95] by working in this downsampled space, building on the signal-processing view of the bilateral filter by Paris and Durand [96] that is mentioned in the IP 03 notes. For enhancement, each grid node stores a color transform. HDRNet [28] stores a 3×43 \times 4 affine matrix per node:

A[i,j,k]∈R3×4,i=0,…,Gy−1,  j=0,…,Gx−1,  k=0,…,Gz−1,A[i, j, k] \in \mathbb{R}^{3 \times 4}, \qquad i = 0, \dots, G_y - 1,\; j = 0, \dots, G_x - 1,\; k = 0, \dots, G_z - 1,

where Gy×GxG_y \times G_x is the number of spatial nodes and GzG_z the number of intensity nodes. To process a full-resolution pixel at row yy and column xx, three steps follow.

1. Guide. Compute a scalar guide value g(x)∈[0,1]g(\mathbf{x}) \in [0, 1] at full resolution, for example the luma, or (as in HDRNet) a learned per-pixel function of the RGB value.

2. Slice. Map the pixel to continuous grid coordinates (uy,ux,uz)=(y sy,  x sx,  g(x) (Gz−1))(u_y, u_x, u_z) = \big(y\, s_y,\; x\, s_x,\; g(\mathbf{x})\,(G_z - 1)\big), where sys_y and sxs_x convert pixel coordinates to grid coordinates, and interpolate the grid there:

Aˉ(x)=∑i,j,kτ(uy−i) τ(ux−j) τ(uz−k)  A[i,j,k],τ(t)=max⁡(1−∣t∣,0),\bar{A}(\mathbf{x}) = \sum_{i, j, k} \tau(u_y - i)\, \tau(u_x - j)\, \tau(u_z - k)\; A[i, j, k], \qquad \tau(t) = \max(1 - |t|, 0),

where τ\tau is the “tent” (linear interpolation) weight. Only the 8 nodes around the pixel have non-zero weight, so this is a trilinear interpolation in (row, column, guide). HDRNet calls this operation slicing; it has no learned parameters and is differentiable with respect to both the grid and the guide [28].

3. Apply. Multiply the pixel’s color by its interpolated matrix:

O(x)=Aˉ(x)[I(x)1],O(\mathbf{x}) = \bar{A}(\mathbf{x}) \begin{bmatrix} I(\mathbf{x}) \\ 1 \end{bmatrix},

where I(x)I(\mathbf{x}) is the RGB column vector and the appended 1 lets the fourth column of Aˉ\bar{A} act as an offset. Each output channel is thus an affine combination of the three input channels.

Left: the luma of one image row plotted against x position, over a grid of 13 vertical lines (spatial nodes) and 8 horizontal lines (intensity nodes); one pixel at x 330 with luma 0.38 is marked with a star, and the cell around it is shaded with its four corner nodes marked. Right: a block diagram. Full-resolution input goes to a downsample box and then to a small CNN or least-squares box that outputs a grid of 3 by 4 affine maps. The full-resolution input also produces a guide map g. A slice box takes the grid and the guide and does a trilinear lookup at x, y, g; an apply box multiplies A by r, g, b, 1; the result is the full-resolution output.
Figure 11 — A bilateral grid of affine color transforms. Left: in one image row, each pixel is located by its position and its guide value, so pixels on the two sides of an edge fall into different intensity cells. Right: the expensive part (the network or the fit) runs on a thumbnail; slicing and applying cost a fixed, small number of operations per full-resolution pixel.

Why the intensity axis matters. If the grid had only spatial axes, neighboring pixels would always get nearly the same transform, and a strong edge inside a cell would be blurred into a halo (a bright fringe on the dark side or a dark fringe on the bright side). The intensity axis separates them: a dark pixel and a bright pixel at the same position read different nodes. This is the same “only average with pixels that look alike” idea as the bilateral filter.

HDRNet. Gharbi et al. [28] train a CNN on a fixed 256 × 256 copy of the image, with a local path and a global path (fully connected layers that summarize the whole image), that outputs a 16 × 16 × 8 grid with 12 coefficients per node; the guide is a learned pointwise function of the RGB value, and the loss is computed on the full-resolution output, so the grid is learned to look right after slicing. The authors report a real-time viewfinder at 1080p on a smartphone [28].

Optional: Paper guide: HDRNet (in the Paper guides). The architecture, the training data, and what to check in its tables.

Bilateral guided upsampling. The same machinery can accelerate any slow operator, learned or not. Chen et al. [79] run the operator on a downsampled image, fit a bilateral grid of affine transforms that maps the downsampled input to the downsampled output, and slice that grid with the full-resolution input. The code below does exactly this, in a simplified form.

The deep guided filter. Wu et al. [97] do the same with the guided filter [80] (see the IP 03 note) as a differentiable layer: it fits the filter’s local linear coefficients between a low-resolution guide and a network’s low-resolution output, upsamples the coefficient maps bilinearly, and applies them to the full-resolution guide. It has no intensity axis, so it relies on small windows rather than intensity bins to keep edges.

Learnable 3D lookup tables

Plain version. A 3D lookup table (3D LUT) is a color cube: for a grid of input colors it stores the output color. Any pixel’s color falls inside one small box of the cube, and its output is blended from the box’s 8 corners. Fixed, hand-tuned 3D LUTs are widely used for color and tone adjustment in camera imaging pipelines and photo editing tools [29]; the learned version lets a small network choose the table for each photo.

Precise version. A 3D LUT with MM nodes per axis is an array Φ∈RM×M×M×3\Phi \in \mathbb{R}^{M \times M \times M \times 3}; node (i,j,k)(i, j, k) holds the output color for the input color (iM−1,jM−1,kM−1)\big(\tfrac{i}{M-1}, \tfrac{j}{M-1}, \tfrac{k}{M-1}\big). For an input color (r,g,b)∈[0,1]3(r, g, b) \in [0, 1]^3, compute the continuous lattice coordinates and split them into an integer corner and a fractional offset:

ur=r (M−1),i=min⁡(⌊ur⌋, M−2),dr=ur−i,u_r = r\,(M - 1), \quad i = \min\big(\lfloor u_r \rfloor,\, M - 2\big), \quad d_r = u_r - i,

and likewise j,dgj, d_g for green and k,dbk, d_b for blue, where ⌊⋅⌋\lfloor \cdot \rfloor rounds down, the cap M−2M - 2 keeps the upper corner i+1i + 1 inside the table when r=1r = 1, and dr,dg,db∈[0,1]d_r, d_g, d_b \in [0, 1] are the offsets inside the cell. Trilinear interpolation blends the 8 corners of the cell:

Φ(r,g,b)=∑a,b′,c∈{0,1}wa(dr) wb′(dg) wc(db)  Φ[i+a, j+b′, k+c],w0(d)=1−d,  w1(d)=d,\Phi(r, g, b) = \sum_{a, b', c \in \{0, 1\}} w_a(d_r)\, w_{b'}(d_g)\, w_c(d_b)\; \Phi[i + a,\, j + b',\, k + c], \qquad w_0(d) = 1 - d,\; w_1(d) = d,

where a,b′,ca, b', c select the lower (0) or upper (1) corner along each axis. The 8 weights are non-negative and sum to 1, so the output is a weighted average of the 8 stored colors. It is the same formula as slicing a bilateral grid, with the axes being (R, G, B) instead of (row, column, intensity).

Three panels. Left: a 5 by 5 by 5 lattice of colored dots in an R, G, B cube, with one input color shown as a red star and lines to the 8 black-circled corners of its cell. Middle: the cell drawn as two square faces, one for the lower and one for the upper blue index, each with four labeled corners and the star at the same relative position, with offsets dr and db marked, and the rule that a corner's weight is the product of d or 1 minus d along each axis. Right: the curve x to the power 1 over 2.2 and its piecewise-linear approximations from 5 and 9 nodes; both are close except near zero, where the true curve is much steeper than the first linear piece.
Figure 13 — A 3D LUT and trilinear interpolation. (a) The lattice; an input color is located in one cell. (b) The 8 corner weights are products of the offsets. (c) Along one axis, interpolation is piecewise linear, so a curve that bends sharply between nodes (here, a gamma curve near black) is approximated poorly unless the lattice is fine there.

Why LUTs are fast. Applying a LUT costs 8 table reads and about 24 multiply–adds per pixel, regardless of how complicated the color transform is. A transform built from dozens of color operations, or one that no formula describes, costs the same as the identity once it is baked into the table. The work is a few memory reads and multiply–adds per pixel with no data-dependent branching, which suits GPUs and fixed-function hardware well. The price is memory and smoothness: the table has 3M33M^3 values, and it can only represent transforms that are close to linear inside each cell.

Image-adaptive 3D LUTs. Zeng et al. [29] made LUTs learnable and content-dependent. They learn NN basis LUTs Φ1,…,ΦN\Phi_1, \dots, \Phi_N and a small CNN that looks at a 256 × 256 downsampled copy of the image and predicts weights w1,…,wNw_1, \dots, w_N. The image’s own LUT is the weighted sum

Φ=∑n=1Nwn Φn,\Phi = \sum_{n=1}^{N} w_n\, \Phi_n,

which is applied to the full-resolution image with one trilinear interpolation per pixel. Because the interpolation is linear in the table entries, fusing the tables first and interpolating once gives the same result as interpolating each table and fusing the outputs. In the paper, M=33M = 33 and N=3N = 3; each LUT has 3⋅333=107,8113 \cdot 33^3 = 107{,}811 values, the whole model has fewer than 600K parameters, and it processes a 4K image in under 2 ms on one Titan RTX GPU [29]. Training adds two regularizers on the tables: a smoothness term that penalizes differences between neighboring nodes, and a monotonicity term that penalizes outputs that decrease when an input channel increases, as a soft preference for componentwise ordering; signed basis weights and finite penalties do not guarantee global brightness ordering. The authors also state the limits clearly: one LUT per image applies the same color transform everywhere, so local adjustments are impossible, and since each color is transformed independently, noise in a night image is amplified rather than removed [29].

Optional: A bilateral-grid experiment and Implementing and checking trilinear LUTs (in the Supplementary notes). Complete tested examples, including the limits of coarse grids and dark-color interpolation.

References

  1. Shyandram, “簡介Deep Learning Low-Light Image Enhancement (LLIE)” (Introduction to deep-learning low-light image enhancement, in Chinese), Medium post, 2024; updated 2026. link
  2. A. Foi, M. Trimeche, V. Katkovnik and K. Egiazarian, “Practical Poissonian-Gaussian noise modeling and fitting for single-image raw-data,” IEEE Transactions on Image Processing, vol. 17, no. 10, pp. 1737–1754, 2008. doi
  3. K. He, J. Sun and X. Tang, “Single image haze removal using dark channel prior,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 33, no. 12, pp. 2341–2353, 2011. doi
  4. X. Guo, Y. Li and H. Ling, “LIME: Low-light image enhancement via illumination map estimation,” IEEE Transactions on Image Processing, vol. 26, no. 2, pp. 982–993, 2017. doi
  5. ITU-R, “Recommendation BT.601-7: Studio encoding parameters of digital television for standard 4:3 and wide-screen 16:9 aspect ratios,” 2011. ITU
  6. Q. Yan, Y. Feng, C. Zhang, et al., “HVI: A new color space for low-light image enhancement,” in Proc. IEEE/CVF CVPR, 2025. arXiv
  7. S. M. Pizer, E. P. Amburn, J. D. Austin, et al., “Adaptive histogram equalization and its variations,” Computer Vision, Graphics, and Image Processing, vol. 39, no. 3, pp. 355–368, 1987. doi
  8. K. Zuiderveld, “Contrast limited adaptive histogram equalization,” in Graphics Gems IV, Academic Press, 1994, pp. 474–485. doi
  9. E. H. Land, “The retinex theory of color vision,” Scientific American, vol. 237, no. 6, pp. 108–128, 1977. doi
  10. D. J. Jobson, Z. Rahman and G. A. Woodell, “Properties and performance of a center/surround retinex,” IEEE Transactions on Image Processing, vol. 6, pp. 451–462, 1997. doi
  11. D. J. Jobson, Z. Rahman and G. A. Woodell, “A multiscale retinex for bridging the gap between color images and the human observation of scenes,” IEEE Transactions on Image Processing, vol. 6, no. 7, pp. 965–976, 1997. doi
  12. X. Dong, Y. Pang and J. Wen, “Fast efficient algorithm for enhancement of low lighting video,” ACM SIGGRAPH 2010 Posters, 2010. SIGGRAPH history doi
  13. S. Wang, J. Zheng, H.-M. Hu and B. Li, “Naturalness preserved enhancement algorithm for non-uniform illumination images,” IEEE Transactions on Image Processing, vol. 22, no. 9, pp. 3538–3548, 2013. doi
  14. C. Lee, C. Lee and C.-S. Kim, “Contrast enhancement based on layered difference representation of 2D histograms,” IEEE Transactions on Image Processing, vol. 22, no. 12, pp. 5372–5384, 2013. doi
  15. C. Li, C. Guo, L. Han, J. Jiang, M.-M. Cheng, J. Gu and C. C. Loy, “Low-light image and video enhancement using deep learning: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 12, pp. 9396–9416, 2022. arXiv
  16. C. Wei, W. Wang, W. Yang and J. Liu, “Deep retinex decomposition for low-light enhancement,” in Proc. British Machine Vision Conference (BMVC), 2018. arXiv project
  17. Y. Jiang, X. Gong, D. Liu, et al., “EnlightenGAN: Deep light enhancement without paired supervision,” IEEE Transactions on Image Processing, vol. 30, pp. 2340–2349, 2021. arXiv
  18. C. Guo, C. Li, J. Guo, et al., “Zero-reference deep curve estimation for low-light image enhancement,” in Proc. IEEE/CVF CVPR, 2020, pp. 1777–1786. arXiv
  19. C. Li, C. Guo and C. C. Loy, “Learning to enhance low-light image via zero-reference deep curve estimation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 8, 2022. arXiv
  20. R. Liu, L. Ma, J. Zhang, X. Fan and Z. Luo, “Retinex-inspired unrolling with cooperative prior architecture search for low-light image enhancement,” in Proc. IEEE/CVF CVPR, 2021, pp. 10556–10565. arXiv
  21. L. Ma, T. Ma, R. Liu, X. Fan and Z. Luo, “Toward fast, flexible, and robust low-light image enhancement,” in Proc. IEEE/CVF CVPR, 2022, pp. 5627–5636. arXiv
  22. Y. Cai, H. Bian, J. Lin, H. Wang, R. Timofte and Y. Zhang, “Retinexformer: One-stage Retinex-based transformer for low-light image enhancement,” in Proc. IEEE/CVF ICCV, 2023. arXiv code
  23. Y. Wang, R. Wan, W. Yang, H. Li, L.-P. Chau and A. C. Kot, “Low-light image enhancement with normalizing flow,” in Proc. AAAI Conference on Artificial Intelligence, vol. 36, no. 3, pp. 2604–2612, 2022. arXiv
  24. X. Yi, H. Xu, H. Zhang, L. Tang and J. Ma, “Diff-Retinex: Rethinking low-light image enhancement with a generative diffusion model,” in Proc. IEEE/CVF ICCV, 2023. arXiv
  25. J. Hou, Z. Zhu, J. Hou, H. Liu, H. Zeng and H. Yuan, “Global structure-aware diffusion process for low-light image enhancement,” in Advances in Neural Information Processing Systems (NeurIPS), 2023. arXiv
  26. H. Jiang, A. Luo, X. Liu, S. Han and S. Liu, “LightenDiffusion: Unsupervised low-light image enhancement with latent-Retinex diffusion models,” in Proc. European Conference on Computer Vision (ECCV), 2024. arXiv
  27. C. He, C. Fang, Y. Zhang, et al., “Reti-Diff: Illumination degradation image restoration with Retinex-based latent diffusion model,” in Proc. International Conference on Learning Representations (ICLR), 2025. proceedings
  28. M. Gharbi, J. Chen, J. T. Barron, S. W. Hasinoff and F. Durand, “Deep bilateral learning for real-time image enhancement,” ACM Transactions on Graphics (Proc. SIGGRAPH), vol. 36, no. 4, art. 118, 2017. doi arXiv
  29. H. Zeng, J. Cai, L. Li, Z. Cao and L. Zhang, “Learning image-adaptive 3D lookup tables for high performance photo enhancement in real-time,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 4, 2022. doi arXiv
  30. K. G. Lore, A. Akintayo and S. Sarkar, “LLNet: A deep autoencoder approach to natural low-light image enhancement,” Pattern Recognition, vol. 61, pp. 650–662, 2017. doi
  31. L. Tao, C. Zhu, J. Song, T. Lu, H. Jia and X. Xie, “Low-light image enhancement using CNN and bright channel prior,” in Proc. IEEE International Conference on Image Processing (ICIP), 2017, pp. 3215–3219. doi
  32. S. W. Zamir, A. Arora, S. Khan, et al., “Learning enriched features for real image restoration and enhancement,” in Proc. European Conference on Computer Vision (ECCV), 2020, pp. 492–511. arXiv
  33. X. Xu, R. Wang, C.-W. Fu and J. Jia, “SNR-aware low-light image enhancement,” in Proc. IEEE/CVF CVPR, 2022, pp. 17693–17703. doi
  34. Z. Cui, K. Li, L. Gu, S. Su, P. Gao, Z. Jiang, Y. Qiao and T. Harada, “You only need 90K parameters to adapt light: A light weight transformer for image enhancement and exposure correction,” in Proc. British Machine Vision Conference (BMVC), 2022. paper arXiv
  35. X. Ding, X. Zhang, N. Ma, J. Han, G. Ding and J. Sun, “RepVGG: Making VGG-style ConvNets great again,” in Proc. IEEE/CVF CVPR, 2021, pp. 13728–13737. doi arXiv
  36. Y. Zhang, J. Zhang and X. Guo, “Kindling the darkness: A practical low-light image enhancer,” in Proc. ACM International Conference on Multimedia (ACM MM), 2019, pp. 1632–1640. doi
  37. Y. Zhang, X. Guo, J. Ma, W. Liu and J. Zhang, “Beyond brightening low-light images,” International Journal of Computer Vision, vol. 129, no. 4, pp. 1013–1037, 2021. doi
  38. J. Bai, Y. Yin, Q. He, Y. Li and X. Zhang, “RetinexMamba: Retinex-based Mamba for low-light image enhancement,” arXiv:2405.03349, 2024. arXiv
  39. S.-E. Weng, S.-G. Miaou and R. Christanto, “A lightweight low-light image enhancement network via channel prior and gamma correction,” International Journal of Pattern Recognition and Artificial Intelligence, vol. 39, no. 12, art. 2554013, 2025. doi code manuscript
  40. S.-E. Weng, C.-Y. Hsiao, L.-W. Lu, et al., “Rethinking theoretical illumination for efficient low-light image enhancement” (CPGA-Net+), arXiv:2409.05274, 2024. arXiv
  41. W. Wu, J. Weng, P. Zhang, X. Wang, W. Yang and J. Jiang, “URetinex-Net: Retinex-based deep unfolding network for low-light image enhancement,” in Proc. IEEE/CVF CVPR, 2022, pp. 5891–5900. doi
  42. Z. Liang, C. Li, S. Zhou, R. Feng and C. C. Loy, “Iterative prompt learning for unsupervised backlit image enhancement,” in Proc. IEEE/CVF ICCV, 2023. arXiv
  43. C. Yang, M. Jin, X. Jia, Y. Xu and Y. Chen, “AdaInt: Learning adaptive intervals for 3D lookup tables on real-time image enhancement,” in Proc. IEEE/CVF CVPR, 2022, pp. 17522–17531. CVF arXiv
  44. C. Yang, M. Jin, Y. Xu, R. Zhang, Y. Chen and H. Liu, “SepLUT: Separable image-adaptive lookup tables for real-time image enhancement,” in Proc. European Conference on Computer Vision (ECCV), 2022, pp. 201–217. doi arXiv
  45. T. Wang, Y. Li, J. Peng, Y. Ma, X. Wang, F. Song and Y. Yan, “Real-time image enhancer via learnable spatial-aware 3D lookup tables,” in Proc. IEEE/CVF ICCV, 2021, pp. 2471–2480. CVF arXiv
  46. C. Liu, H. Yang, J. Fu and X. Qian, “4D LUT: Learnable context-aware 4D lookup table for image enhancement,” IEEE Transactions on Image Processing, vol. 32, pp. 4742–4756, 2023. doi arXiv
  47. W. Kim and N. I. Cho, “Image-adaptive 3D lookup tables for real-time image enhancement with bilateral grids,” in Proc. European Conference on Computer Vision (ECCV), 2024, pp. 91–108. doi ECCV
  48. C. Chen, Q. Chen, J. Xu and V. Koltun, “Learning to see in the dark,” in Proc. IEEE/CVF CVPR, 2018, pp. 3291–3300. arXiv
  49. M. V. Conde, G. Geigle and R. Timofte, “InstructIR: High-quality image restoration following human instructions,” in Proc. European Conference on Computer Vision (ECCV), 2024. arXiv
  50. Z. Cui, G.-J. Qi, L. Gu, S. You, Z. Zhang and T. Harada, “Multitask AET with orthogonal tangent regularity for dark object detection,” in Proc. IEEE/CVF ICCV, 2021. arXiv
  51. K. A. Hashmi, G. Kallempudi, D. Stricker and M. Z. Afzal, “FeatEnHancer: Enhancing hierarchical features for object detection and beyond under low-light vision,” in Proc. IEEE/CVF ICCV, 2023. arXiv
  52. M. Dehghani, A. Arnab, L. Beyer, A. Vaswani and Y. Tay, “The efficiency misnomer,” in Proc. International Conference on Learning Representations (ICLR), 2022. ICLR arXiv
  53. A. Ignatov, R. Timofte, S. Liu, et al., “Learned smartphone ISP on mobile GPUs with deep learning, Mobile AI & AIM 2022 challenge: Report,” in Proc. European Conference on Computer Vision (ECCV) Workshops, 2023, pp. 44–70. doi arXiv
  54. Z. Wang, A. C. Bovik, H. R. Sheikh and E. P. Simoncelli, “Image quality assessment: From error visibility to structural similarity,” IEEE Transactions on Image Processing, vol. 13, no. 4, pp. 600–612, 2004. doi
  55. R. Zhang, P. Isola, A. A. Efros, E. Shechtman and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” in Proc. IEEE/CVF CVPR, 2018. arXiv
  56. K. Ma, K. Zeng and Z. Wang, “Perceptual quality assessment for multi-exposure image fusion,” IEEE Transactions on Image Processing, vol. 24, no. 11, pp. 3345–3356, 2015. doi
  57. V. Vonikakis, “Datasets” (source of the VV test images), personal web page. link
  58. A. Mittal, R. Soundararajan and A. C. Bovik, “Making a ‘completely blind’ image quality analyzer,” IEEE Signal Processing Letters, vol. 20, no. 3, pp. 209–212, 2013. doi
  59. A. Mittal, A. K. Moorthy and A. C. Bovik, “No-reference image quality assessment in the spatial domain,” IEEE Transactions on Image Processing, vol. 21, no. 12, pp. 4695–4708, 2012. doi
  60. Y. Blau, R. Mechrez, R. Timofte, T. Michaeli and L. Zelnik-Manor, “The 2018 PIRM challenge on perceptual image super-resolution,” in Proc. European Conference on Computer Vision (ECCV) Workshops, 2018. arXiv
  61. C. Ma, C.-Y. Yang, X. Yang and M.-H. Yang, “Learning a no-reference quality metric for single-image super-resolution,” Computer Vision and Image Understanding, vol. 158, pp. 1–16, 2017. doi arXiv
  62. D. Hasler and S. E. Süsstrunk, “Measuring colorfulness in natural images,” in Proc. SPIE 5007, Human Vision and Electronic Imaging VIII, 2003. doi
  63. J. Immerkær, “Fast noise variance estimation,” Computer Vision and Image Understanding, vol. 64, no. 2, pp. 300–302, 1996. doi
  64. J. Liao, S. Hao, R. Hong and M. Wang, “GT-Mean loss: A simple yet effective solution for brightness mismatch in low-light image enhancement,” in Proc. IEEE/CVF ICCV, 2025. arXiv
  65. Y. P. Loh and C. S. Chan, “Getting to know low-light images with the Exclusively Dark dataset,” Computer Vision and Image Understanding, vol. 178, pp. 30–42, 2019. arXiv
  66. F. Liu and L. Fan, “A review of advancements in low-light image enhancement using deep learning,” Neurocomputing, vol. 652, art. 131052, 2025. arXiv
  67. J. Liu, D. Xu, W. Yang, M. Fan and H. Huang, “Benchmarking low-light image enhancement and beyond,” International Journal of Computer Vision, vol. 129, no. 4, pp. 1153–1184, 2021. doi project
  68. W. Yang, W. Wang, H. Huang, S. Wang and J. Liu, “Sparse gradient regularized deep Retinex network for robust low-light image enhancement,” IEEE Transactions on Image Processing, vol. 30, pp. 2072–2086, 2021. doi
  69. J. Hai, Z. Xuan, R. Yang, et al., “R2RNet: Low-light image enhancement via Real-low to Real-normal Network,” Journal of Visual Communication and Image Representation, vol. 90, art. 103712, 2023. arXiv
  70. R. Wang, X. Xu, C.-W. Fu, et al., “Seeing dynamic scene in the dark: A high-quality video dataset with mechatronic alignment,” in Proc. IEEE/CVF ICCV, 2021, pp. 9680–9689. doi code
  71. V. Bychkovsky, S. Paris, E. Chan and F. Durand, “Learning photographic global tonal adjustment with a database of input/output image pairs,” in Proc. IEEE CVPR, 2011, pp. 97–104. dataset
  72. J. Cai, S. Gu and L. Zhang, “Learning a deep single image contrast enhancer from multi-exposure images,” IEEE Transactions on Image Processing, vol. 27, no. 4, pp. 2049–2062, 2018. doi
  73. X. Liu, Z. Wu, A. Li, et al., “NTIRE 2024 challenge on low light image enhancement: Methods and results,” in Proc. IEEE/CVF CVPR Workshops, 2024. arXiv
  74. X. Liu, Z. Wu, F.-A. Vasluianu, et al., “NTIRE 2025 challenge on low light image enhancement: Methods and results,” in Proc. IEEE/CVF CVPR Workshops, 2025. arXiv
  75. S. Zheng, Y. Ma, J. Pan, C. Lu and G. Gupta, “Low-light image and video enhancement: A comprehensive survey and beyond,” arXiv:2212.10772, 2022. arXiv
  76. E. Adhikarla, Y. Liu and B. D. Davison, “Diffusion models for low-light image enhancement: A multi-perspective taxonomy and performance analysis,” arXiv:2510.05976, 2025. arXiv
  77. J. Yan, C. Tu, W. Zhang, Z. Wang, P. Cao, Q. Lin, Y. Fang, X. Liu, Z. Wu, Z. Zhou and R. Timofte, “NTIRE 2026 challenge on efficient low light image enhancement: Methods and results,” NTIRE 2026 workshop report, arXiv:2605.02212, 2026. arXiv
  78. O. Ronneberger, P. Fischer and T. Brox, “U-Net: Convolutional networks for biomedical image segmentation,” in Proc. Medical Image Computing and Computer-Assisted Intervention (MICCAI), 2015, pp. 234–241. doi arXiv
  79. J. Chen, A. Adams, N. Wadhwa and S. W. Hasinoff, “Bilateral guided upsampling,” ACM Transactions on Graphics (Proc. SIGGRAPH Asia), vol. 35, no. 6, 2016. doi project
  80. K. He, J. Sun and X. Tang, “Guided image filtering,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 6, pp. 1397–1409, 2013. doi
  81. C. Chen and J. Mo, “IQA-PyTorch: PyTorch toolbox for image quality assessment,” software (pip package pyiqa). GitHub
  82. J. Ke, Q. Wang, Y. Wang, P. Milanfar and F. Yang, “MUSIQ: Multi-scale image quality transformer,” in Proc. IEEE/CVF ICCV, 2021. arXiv
  83. J. Wang, K. C. K. Chan and C. C. Loy, “Exploring CLIP for assessing the look and feel of images,” in Proc. AAAI Conference on Artificial Intelligence, 2023. arXiv
  84. V. Kannan, S. Malik, N. C. Babu and R. Soundararajan, “Quality assessment of low-light restored images: A subjective study and an unsupervised model,” IEEE Access, vol. 11, pp. 68216–68230, 2023. doi arXiv
  85. ITU-R, “Recommendation BT.500-15: Methodologies for the subjective assessment of the quality of television images,” 2023. ITU
  86. R. A. Bradley and M. E. Terry, “Rank analysis of incomplete block designs: The method of paired comparisons,” Biometrika, vol. 39, no. 3–4, pp. 324–345, 1952. doi
  87. H. Talebi and P. Milanfar, “NIMA: Neural image assessment,” IEEE Transactions on Image Processing, 2018. arXiv
  88. S. Yang, T. Wu, S. Shi, et al., “MANIQA: Multi-dimension attention network for no-reference image quality assessment,” in Proc. IEEE/CVF CVPR Workshops, 2022, pp. 1190–1199. doi arXiv
  89. H. Wu, Z. Zhang, W. Zhang, C. Chen, et al., “Q-Align: Teaching LMMs for visual scoring via discrete text-defined levels,” in Proc. International Conference on Machine Learning (ICML), PMLR 235, 2024. arXiv
  90. A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto and H. Adam, “MobileNets: Efficient convolutional neural networks for mobile vision applications,” arXiv:1704.04861, 2017. arXiv
  91. M. Sandler, A. Howard, M. Zhu, A. Zhmoginov and L.-C. Chen, “MobileNetV2: Inverted residuals and linear bottlenecks,” in Proc. IEEE/CVF CVPR, 2018, pp. 4510–4520. doi arXiv
  92. W.-S. Lai, J.-B. Huang, O. Wang, E. Shechtman, E. Yumer and M.-H. Yang, “Learning blind video temporal consistency,” in Proc. European Conference on Computer Vision (ECCV), 2018, pp. 179–195. doi arXiv
  93. S. W. Hasinoff, D. Sharlet, R. Geiss, A. Adams, J. T. Barron, F. Kainz, J. Chen and M. Levoy, “Burst photography for high dynamic range and low-light imaging on mobile cameras,” ACM Transactions on Graphics (Proc. SIGGRAPH Asia), vol. 35, no. 6, 2016. doi
  94. J. Chen, S. Paris and F. Durand, “Real-time edge-aware image processing with the bilateral grid,” ACM Transactions on Graphics (Proc. SIGGRAPH), vol. 26, no. 3, art. 103, 2007. doi
  95. C. Tomasi and R. Manduchi, “Bilateral filtering for gray and color images,” in Proc. IEEE International Conference on Computer Vision (ICCV), 1998, pp. 839–846. doi
  96. S. Paris and F. Durand, “A fast approximation of the bilateral filter using a signal processing approach,” International Journal of Computer Vision, vol. 81, no. 1, pp. 24–52, 2009. doi
  97. H. Wu, S. Zheng, J. Zhang and K. Huang, “Fast end-to-end trainable guided filter,” in Proc. IEEE/CVF CVPR, 2018, pp. 1838–1847. doi arXiv
  98. R. Wang, Q. Zhang, C.-W. Fu, X. Shen, W.-S. Zheng and J. Jia, “Underexposed photo enhancement using deep illumination estimation,” in Proc. IEEE/CVF CVPR, 2019, pp. 6842–6850. doi CVF
  99. Y. Lin, Z. Fu, K. Wen, T. Ye, S. Chen, G. Meng, Y. Wang, et al., “DPLUT: Unsupervised low-light image enhancement with lookup tables and diffusion priors,” in Proc. AAAI Conference on Artificial Intelligence, vol. 39, no. 5, pp. 5316–5324, 2025. doi arXiv
  100. K. Dabov, A. Foi, V. Katkovnik and K. Egiazarian, “Image denoising by sparse 3-D transform-domain collaborative filtering,” IEEE Transactions on Image Processing, 2007. doi
  101. K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” ICLR, 2015. arXiv
  102. J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li and L. Fei-Fei, “ImageNet: A large-scale hierarchical image database,” CVPR, 2009. doi
  103. A. Brateanu, T. Mu, C. O. Ancuti and C. Ancuti, “Multinex: Lightweight Low-light Image Enhancement via Multi-prior Retinex,” in Proc. IEEE/CVF CVPR, 2026. paper supplement