Deep Retinex Decomposition for Low-Light Enhancement
2018 BMVC
这篇文章是我前后读过许多遍,比较值得介绍。受Retinex理论的启发,它采用了两阶段式的先分解后增强的步骤,完全采用CNN实现。对于Decom-Net的训练,引入了反射图一致性约束(consistency of reflectance)和光照图平滑性约束(smoothness of illumination),非常容易复现,实验效果也不错。
[1] Shi, W., Loy, C. C., & Tang, X. (2016). Deep Specialized Network for Illuminant Estimation. ECCV, 9908, 371–387. doi.org/10.1007/978-3-3
[2] Guo, X., Li, Y., & Ling, H. (2017). LIME: Low-light image enhancement via illumination map estimation. IEEE Transactions on Image Processing, 26(2), 982–993. doi.org/10.1109/TIP.201
[3] Lore, K. G., Akintayo, A., & Sarkar, S. (2017). LLNet: A deep autoencoder approach to natural low-light image enhancement. Pattern Recognition, 61, 650–662. doi.org/10.1016/j.patco
[4] Shen, L., Yue, Z., Feng, F., Chen, Q., Liu, S., & Ma, J. (2017). MSR-net:Low-light Image Enhancement Using Deep Convolutional Network. ArXiv. Retrieved from arxiv.org/abs/1711.0248
[5] Cai, J., Gu, S., & Zhang, L. (2018). Learning a Deep Single Image Contrast Enhancer from Multi-Exposure Images. IEEE Transactions on Image Processing, 1(c), 1–14. Retrieved from www4.comp.polyu.edu.hk/
[6] Gao, Y., Hu, H. M., Li, B., & Guo, Q. (2018). Naturalness preserved nonuniform illumination estimation for image enhancement based on retinex. IEEE Transactions on Multimedia, 20(2), 335–344. doi.org/10.1109/TMM.201
[7] Chen, C., Chen, Q., Xu, J., & Koltun, V. (2018). Learning to See in the Dark. CVPR, 3291–3300. doi.org/10.1109/CVPR.20
[8] Lv, F., Lu, F., Wu, J., & Lim, C. (2018). MBLLEN: Low-light Image/Video Enhancement Using CNNs. BMVC, 1–13.
[9] Wei, C., Wang, W., Yang, W., & Liu, J. (2018). Deep Retinex Decomposition for Low-Light Enhancement. BMVC, (61772043). Retrieved from arxiv.org/abs/1808.0456
[10] Zhang, Y., Zhang, J., & Guo, X. (2019). Kindling the Darkness: A Practical Low-light Image Enhancer. ArXiv, 1–13. Retrieved from arxiv.org/abs/1905.0416