在拍照時時常會出現手震不穩等情況使得影像產生模糊。在影像處理的問題上，圖像去模糊一直是一個重要的課題。隨著深度學習的發展，越來越多方法提出使用卷積網路(CNN) 的架構來生成清晰的圖像。為了解決此問題，我們提出了一種基於卷積網路的多補丁注意力網絡（Deep Multi-patch Attention Network , DMPAN）來還原清晰的圖像。我們將圖片切割成多個區塊，並且使用多層網路架構對模糊的圖像萃取特徵。我們還在網絡中添加注意模塊，透過此機制讓網路能得到更有意義的特徵且更加專注於還原局部的細節區塊。DMPAN在PSNR和SSIM上與現有方法比較表現出色，對於圖像的細節還原能力更好且維持少量的網路參數。

When taking pictures, there are often occur handshake and other situations that make the image blurry. On the issue of image processing, image deblurring has been an important issue. With the development of deep learning, more and more methods have proposed convolutional network (CNN) architecture to generate clear images. To solve this problem, we propose a CNN based deep multi-patch attention network (DMPAN) to restore clear images. We cut the picture into multiple patches, and use a multi-layer network architecture to extract features from the blurred image. We also add attention modules to the network, through which the network can get more meaningful features and focus more on restoring details. Compared with existing methods, DMPAN performs well on PSNR and SSIM, and has better ability to restore the details of the image and maintain a small amount of parameters

[1] S. Nah, T. Kim, and K.M Lee. Deep multi-scale convolutional neural network for dynamic scene deblurring. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 257 – 265, 2017.
[2] X. Tao, H. Gao, X. Shen, J. Wang, and J. Jia. Scale-recurrent network for deep image deblurring. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 8174–8182, 2018.
[3] J. Zhang, J. Pan, J. Ren, Y. Song, L. Bao, R. Lau, and M. Yang. Dynamic scene deblurring using spatially variant recurrent neural networks. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 2521–2529, 2018.
[4] A. Krizhevsky, I. Sutskever, and G.E Hinton. Imagenet classification with deep convolutional neural networks. In Proc. Adv. Neural Inf. Process. Syst, pages 1097–1105, 2012.
[5] K. Simonyan and A. Zisserman. Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556, 2014.
[6] K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. In CVPR, pages 770–778. IEEE, 2016.
[7] J. Hu, L. Shen, and G. Sun. Squeeze-and-excitation networks. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 7132–7141, 2018.
[8] S. Woo, J. Park, J. Lee, and I. Kweon. Cbam: Convolutional block attention module. In Proceedings of the European Conference on Computer Vision (ECCV), pages 3–19, 2018.
[9] J. Kim, J. Choi, M.Cheon, and J. Lee. Ram: Residual attention module for single image super-resolution. arXiv preprint arXiv:1811.12043, 2018.
[10] D.P Kingma and J. Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014.
[11] Huang G, Liu Z, Maaten Laurens VD, W. Kilian Q. Densely Connected Convolutional Networks. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2017:2261-2269.
[12] J. Sun, W. Cao, Z. Xu, and J. Ponce. Learning a convolutional neural network for nonuniform motion blur removal. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 769–777, 2015
[13] K. He, X. Zhang, S. Ren, and JianSun. Spatial pyramid pooling in deep convolutional networks for visual recognition. In Proc. Eur. Conf. Comp. Vis, pages 346–361. Springer, 2014.
[14] P. Koniusz, H. Zhang, and F. Porikli. A deeper look at power normalizations. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 5774–5783, 2018.
[15] S. Lazebnik, C. Schmid, and J. Ponce. Beyond bags of features: Spatial pyramid matching for recognizing natural scene categories. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 2169– 2178. IEEE, 2006.
[16] X. Lu, Z. Lin, X. Shen, R. Mech, and JZ. Wang, “Deep multi-patch aggregation network for image style, aesthetics, and quality estimation,”In Proc. IEEE Conf. Comp. Vis, pages 990–998, 2015.
[17] O. Kupyn, V. Budzan, M. Mykhailych, D. Mishkin, and J. Matas. Deblurgan: Blind motion deblurring using conditional adversarial networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 8183–8192, 2018.
[18] H. Zhang, Y. Dai, H. Li, and P. Koniusz. Deep stacked hierarchical multi-patch network for image deblurring. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 5978– 5986, 2019