基于多级特征融合的低光图像增强网络

doi:10.3778/j.issn.1002-8331.2401-0249

摘要/Abstract

摘要： 近年来，将Retinex理论与CNN结合的低光图像增强方法已取得显著成效，其中KinD++方法在光照调节方面表现尤为突出。然而当处理光照极低或极不均匀图像时该方法仍有提升空间。针对KinD++方法处理光照极低或极不均匀图像时存在的细节模糊、伪影以及色彩失真问题，对分解网和恢复网进行改进，提出一种基于多级特征融合的低光图像增强网络。通过深、中、浅层特征的跨级融合实现了对反射分量更准确的估计；在恢复网中设计De_Block去噪块和Canny边缘增强块，在有效去噪的同时保持细节清晰；采用色彩损失进一步得到自然明亮的图像色彩。实验结果表明，该方法能在有效改善图像亮度的同时修复细节和色彩，避免伪影。相比原方法，在LOL数据集上，PSNR、SSIM以及NIQE分别提高了13%、12%和28%；在多个基准数据集，如LIME、VV、MEF、DICM和LLIV-Phone-imgT上，也展现出了优越的性能。

关键词: 低光图像增强, 非均匀光/极低光照, Retinex理论, 多级特征融合, Canny边缘增强

Abstract: In recent years, the integration of Retinex theory and CNN for low-light image enhancement has achieved sig nificant progress, the KinD++ method stands out particularly in illumination adjustment. However, this method still has shortcomings when dealing with images with uneven or extremely low illumination. In order to address the issues of detail blurring, artifacting, and color distortion, when the KinD++ processes images with uneven or extremely low illumination, this paper makes improvements on decomposition network and restoration network, proposes a low-light image enhancement network of multi-level feature fusion. Through cross-level fusion of deep, medium, and shallow features, a more accurate estimation of the reflection is achieved. In the restoration network, De_Block denoising block and Canny edge enhancement block are designed, which can keep the details clear while denoising effectively. A color loss function is used to achieve natural and bright image colors. The experimental results demonstrate that the proposed method can effectively improve image brightness while restoring details and colors, thereby avoiding artifacts. Compared to the original method, on the LOL dataset, PSNR, SSIM, and NIQE are improved by 13%, 12%, and 28% respectively. Additionally, superior performance is demonstrated on multiple benchmark datasets such as LIME, VV, MEF, DICM, and LLIV-Phone-imgT.

Key words: low-light image enhancement, uneven/extremely low illumination, Retinex theory, multi-level feature fusion, Canny edge enhancement

牟琦, 马悦悦, 李洪安, 李占利. 基于多级特征融合的低光图像增强网络[J]. 计算机工程与应用, 2025, 61(10): 238-246.

MU Qi, MA Yueyue, LI Hong’an, LI Zhanli. Low-Light Image Enhancement Network of Multi-Level Feature Fusion[J]. Computer Engineering and Applications, 2025, 61(10): 238-246.

参考文献

[1] GUO X J, LI Y, LING H B. LIME: low-light image enhancement via illumination map estimation[J]. IEEE Transactions on Image Processing, 2017, 26(2): 982-993.
[2] 彭大鑫, 甄彤, 李智慧. 低光照图像增强研究方法综述[J]. 计算机工程与应用, 2023, 59(18): 14-27.
PENG D X, ZHEN T, LI Z H. Survey of research methods for low light image enhancement[J]. Computer Engineering and Applications, 2023, 59(18): 14-27.
[3] 赵兴运, 孙帮勇. 融合注意力机制和上下文信息的微光图像增强[J]. 中国图象图形学报, 2022, 27(5): 1565-1576.
ZHAO X Y, SUN B Y. An attention mechanism and contextual information based low-light image enhancement method[J]. Journal of Image and Graphics, 2022, 27(5): 1565-1576.
[4] 王炜昊, 王夏黎, 武历展, 等. 融合多尺度密集块的低照度交通图像增强模型[J]. 计算机工程与应用, 2023, 59(17): 223-231.
WANG W H, WANG X L, WU L Z, et al. Low illumination traffic image enhancement model based on multi-scale dense blocks[J]. Computer Engineering and Applications, 2023, 59(17): 223-231.
[5] 朱凯, 李理, 张彤, 等. 视觉Transformer在低级视觉领域的研究综述[J]. 计算机工程与应用, 2024, 60(4): 39-56.
ZHU K, LI L, ZHANG T, et al. Survey of vision transformer in low-level computer vision[J]. Computer Engineering and Applications, 2024, 60(4): 39-56.
[6] IBRAHIM H, PIK KONG N S. Brightness preserving dynamic histogram equalization for image contrast enhancement[J]. IEEE Transactions on Consumer Electronics, 2007, 53(4): 1752-1758.
[7] ABDULLAH-AL-WADUD M, KABIR M H, DEWAN M A A, et al. A dynamic histogram equalization for image contrast enhancement[J]. IEEE Transactions on Consumer Electronics, 2007, 53(2): 593-600.
[8] LI M D, LIU J Y, YANG W H, et al. Structure-revealing low-light image enhancement via robust retinex model[J]. IEEE Transactions on Image Processing, 2018, 27(6): 2828-2841.
[9] GU Z H, LI F, FANG F M, et al. A novel retinex-based fractional-order variational model for images with severely low light[J]. IEEE Transactions on Image Processing, 2019, 29: 3239-3253.
[10] PARK S, YU S, MOON B, et al. Low-light image enhancement using variational optimization-based retinex model[J]. IEEE Transactions on Consumer Electronics, 2017, 63(2): 178-184.
[11] LAND E H. The retinex theory of color vision[J]. Scientific American, 1977, 237(6): 108-128.
[12] LI C Y, GUO C L, HAN L H, et al. Low-light image and video enhancement using deep learning: a survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(12): 9396-9416.
[13] WEI C, WANG W J, YANG W H, et al. Deep Retinex decomposition for low-light enhancement[J]. arXiv:1808. 04560, 2018.
[14] ZHANG Y H, ZHANG J W, GUO X J, et al. Kindling the darkness: a practical low-light image enhancer[C]//Proceedings of the 27th ACM International Conference on Multimedia. New York: ACM, 2019: 1632-1640.
[15] LV X Q, SUN Y J, ZHANG J, et al. Low-light image enhancement via deep Retinex decomposition and bilateral learning[J]. Signal Processing: Image Communication, 2021, 99: 116466.
[16] ZHAO Z J, XIONG B S, WANG L, et al. RetinexDIP: a unified deep framework for low-light image enhancement[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2022, 32(3): 1076-1088.
[17] DABOV K, FOI A, KATKOVNIK V, et al. Image denoising with block-matching and 3D filtering[C]//SPIE 6064: Image Processing: Algorithms and Systems, Neural Networks, and Machine Learning, 2006: 354-365.
[18] RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]//Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2015: 234-241.
[19] ZHANG Y H, GUO X J, MA J Y, et al. Beyond brightening low-light images[J]. International Journal of Computer Vision, 2021, 129(4): 1013-1037.
[20] WANG W J, WEI C, YANG W H, et al. GLADNet: low-light enhancement network with global awareness[C]//Proceedings of the 2018 13th IEEE International Conference on Automatic Face & Gesture Recognition. Piscataway: IEEE, 2018: 751-755.
[21] GUO C L, LI C Y, GUO J C, et al. Zero-reference deep curve estimation for low-light image enhancement[C]//Proceedings of the 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2020: 1777-1786.
[22] WANG Y F, WAN R J, YANG W H, et al. Low-light image enhancement with normalizing flow[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2022, 36(3): 2604-2612.
[23] TANG L F, MA J Y, ZHANG H, et al. DRLIE: flexible low-light image enhancement via disentangled representations[J]. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35(2): 2694-2707.
[24] WU W H, WENG J, ZHANG P P, et al. URetinex-Net: Retinex-based deep unfolding network for low-light image enhancement[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2022: 5891-5900.
[25] HUANG H M, LIN L F, TONG R F, et al. UNet 3+: a full-scale connected UNet for medical image segmentation[C]//Proceedings of the 2020 IEEE International Conference on Acoustics, Speech and Signal Processing. Piscataway: IEEE, 2020: 1055-1059.
[26] HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 770-778.
[27] WOO S, PARK J, LEE J Y, et al. CBAM: convolutional block attention module[C]//Proceedings of the 15th European Conference on Computer Vision. Cham: Springer, 2018: 3-19.
[28] CANNY J. A computational approach to edge detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, 8(6): 679-698.
[29] MITTAL A, SOUNDARARAJAN R, BOVIK A C. Making a “completely blind” image quality analyzer[J]. IEEE Signal Processing Letters, 2013, 20(3): 209-212.
[30] DE BOER J F, CENSE B, PARK B H, et al. Improved signal-to-noise ratio in spectral-domain compared with time-domain optical coherence tomography[J]. Optics Letters, 2003, 28(21): 2067-2069.
[31] WANG Z, BOVIK A C, SHEIKH H R, et al. Image quality assessment: from error visibility to structural similarity[J]. IEEE Transactions on Image Processing, 2004, 13(4): 600-612.
[32] 赵全友, 潘保昌, 郑胜林, 等. 一种颜色保持的彩色图像增强新算法[J]. 计算机应用, 2008, 28(2): 448-451.
ZHAO Q Y, PAN B C, ZHENG S L, et al. New hue preserving algorithm for color image enhancement[J]. Journal of Computer Applications, 2008, 28(2): 448-451.