深度网络图像去雾对比研究

Abstract

目的：比较不同编码—解码结构在浓雾图像恢复中的表现，分析多尺度特征融合与自注意力机制对去雾质量的影响。方法：基于 Dense-Haze 成对图像开展探索性实验，构建 ResNet-50 编码器—解码器、U-Net 及嵌入多头自注意力的 U-Net 模型；统一将图像缩放至 256×256 像素，批量大小为 4，学习率为 1×10-4，训练 30 轮，以均方误差为损失函数，并从收敛趋势、残余雾、结构细节、色彩恢复和伪影等方面进行比较。结果：3 种模型的训练损失均呈下降趋势。ResNet 模型虽收敛较快，但输出存在模糊、亮度不足和棋盘状伪影；U-Net 借助跳跃连接改善了结构与色彩恢复；加入自注意力后，所示样例中的大范围雾分布处理更均衡，整体自然度相对较好。结论：多尺度特征传递是浓雾图像恢复的重要基础，全局依赖建模具有进一步改善视觉一致性的潜力；由于原实验未保留独立测试集及峰值信噪比、结构相似性等完整记录，相关结论仍需通过统一划分、重复训练和外部数据验证。

Keywords

• 图像去雾；U-Net；残差网络；自注意力；Dense-Haze

References

1. [1]He, K., Sun, J. and Tang, X. (2011) Single Image Haze Removal Using Dark Channel Prior. IEEE Transactions on Pattern Analysis and Machine Intelligence, 33, 2341-2353.
2. [2] Cai, B., Xu, X., Jia, K., Qing, C. and Tao, D. (2016) DehazeNet: An End-to-End System for Single Image Haze Removal. IEEE Transactions on Image Processing, 25, 5187-5198.
3. [3]Li, B., Peng, X., Wang, Z., Xu, J. and Feng, D. (2017) AOD-Net: All-In-One Dehazing Network. Proceedings of the IEEE International Conference on Computer Vision, 4770-4778.
4. [4]He, K., Zhang, X., Ren, S. and Sun, J. (2016) Deep Residual Learning for Image Recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,Las Vegas, 770-778.
5. [5]Ronneberger, O., Fischer, P. and Brox, T. (2015) U-Net: Convolutional Networks for Biomedical Image Segmentation. Medical Image Computing and ComputerAssisted Intervention, Munich, 234-241.
6. [6]Vaswani, A., Shazeer, N., Parmar, N., et al. (2017) Attention Is All You Need. Advances in Neural Information Processing Systems, 30, 5998-6008.
7. [7]Ancuti, C.O., Ancuti, C., Sbert, M. and Timofte, R. (2019) Dense Haze: A Benchmark for Image Dehazing with Dense-Haze and Haze-Free Images. Proceedings of the IEEE International Conference on Image Processing, 1014-1018.
8. [8]Liu, X., Ma, Y., Shi, Z. and Chen, J. (2019) GridDehazeNet: Attention-Based Multi-Scale Network for Image Dehazing. Proceedings of the IEEE/CVF International Conference on Computer Vision, 7314-7323.
9. [9]Ancuti, C.O., Ancuti, C., Timofte, R., et al. (2019) NTIRE 2019 Image Dehazing Challenge Report. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, Long Beach, 2241-2253.