隨著深度學習的成熟，在電腦視覺中的去模糊領域也成為發展的主題之一。在處理去模糊的問題上，圖片會運動模糊而造成模糊的現象，近年來也提出很多種深度神經網路架構來去處理這一類的問題。我們提出一種結合超解析領域以及去模糊領域中的想法實現輕量化多補丁神經網路。我們將圖片分割成多塊並由特色聚合的方式去增強去模糊的效果。我們也提供了數組圖片去比較圖片的去模糊程度。LMPAN 在PSNR和SSIM上與現有方法比較上也有優勢，讓生成的圖片可以恢復清晰圖片。

With the maturity of deep learning, the field of deblurring in computer vision has also become one of the development themes. In dealing with the problem of deblurring, the picture will be motion blurred and cause a blurring phenomenon, and many kinds of deep neural network architectures have been proposed to deal with this kind of problem in recent years. We propose a lightweight multi-patch neural network (LMPAN) that combines ideas from the super-resolution and deblurring domains. We partition the image into multiple pieces and enhance the deblurring effect by its feature aggregation. LMPAN that can restore a blurred image to a clear image also has advantages over existing methods in terms of PSNR and SSIM.

[1] S. Nah, K.M Lee, and T. Kim. 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. Jia, and J. Wang. Scale-recurrent network for deep image deblurring. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn, pages 8174 – 8182, 2018
[3] H. Zhang, Y. Dai, P. Koniusz, and H. Li. Deep stacked hierarchical multi-patch network for image deblurring. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5978 – 5986, 2019
[4] Zheng Hui, Xinbo Gao, and Xiumei Wang. Fast and Accurate Single Image Super-Resolution via Information Distillation Network. In IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 723 – 731, 2018
[5] Liu, J., Zhang, W., Tang, J., Tang, Y., Wu, G.: Residual feature aggregation net-
work for image super-resolution. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). pages 2356 - 2365. IEEE, 2020
[6] Jie Liu, Gangshan Wu., and Jie Tang. Residual feature distillation network for lightweight image super-resolution.arXiv preprint (ECCV workshops(3)), pages 41 - 55, 2020
[6] J. Sun, W. Cao, J. Ponce, and Z. Xu. Learning a convolutional neural network for nonuniform motion blur removal. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 769 – 777, 2015
[7] Dong Gong, Jie Yang, Yanning Zhang, Lingqiao Liu, IanReid, Chunhua Shen, Anton Van Den Hengel, and QinfengShi. From motion blur to motion flow: a deep learning solution for removing heterogeneous motion blur. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 2319 – 2328, 2017
[8] Orest Kupyn, Tetiana Martyniuk, Zhangyang Wang, and Junru Wu. DeblurGAN-v2: Deblurring (orders-of magnitude) faster and better. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 1 – 8, 2019.
[9] Gated Fusion Network for Joint Image Deblurring and Super-Resolution 2018
[10] Chao Dong, Chen Change Loy, Xiaoou Tang, and Kaiming He. Learning a deep convolutional network for image super-resolution. In European Conference on Computer Vision (ECCV). pages 184 – 199, 2014
[11] Chao Dong, Chen Change Loy, Xiaoou Tang, and Kaiming He. Image super-resolution using deep convolutional networks. IEEE Transactions on Pattern Analysis and Machine Intelligence 38, 2 (2016), pages 295 – 307, 2016
[12] Chao Dong, Xiaoou Tang, and Chen Change Loy. Accelerating the super-resolution convolutional neural network. In European Conference on Computer Vision (ECCV). pages 391 – 407, 2016
[13] Yulun Zhang, Kunpeng Li, Kai Li, Yun Fu, and Bineng Zhong. Residual
Non-local Attention Networks for Image Restoration. In International Conference
on Learning Representations (ICLR), pages 1 – 8, 2019.
[14] Hui, Z., Gao, X., Yang, Y., Wang, X.: Lightweight image super-resolution with information multi-distillation network. In: ACM Multimedia. Pages 2024 - 2032. ACM ,2019
[15] Namhyuk Ahn, Kyung-Ah Sohn, and Byungkon Kang. Fast, Accurate, and Lightweight Super-Resolution with Cascading Residual Network. In European Conference on Computer Vision (ECCV). pages 252 – 268, 2018
[16] Shuochen Su, Mauricio Delbracio, Guillermo Sapiro, Jue Wang, Wolfgang Heidrich, and Oliver Wang. Deep video deblurring for hand-held cameras. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., volume 2, pages 237 – 246, 2017
[17] Jiawei Zhang, Jinshan Pan, Jimmy Ren, Yibing Song, Linchao Bao, Rynson WH Lau, and Ming-Hsuan Yang. Dynamic scene deblurring using spatially variant recurrent neural networks. In Proc. IEEE Conf. Comp. Vis. Patt. Recogn., pages 2521–2529, 2018
[18] S. Xingjian, Z. Chen, H. Wang, D.-Y. Yeung, W.-K. Wong, and W.-c. Woo. Convolutional lstm network: A machine learning approach for precipitation nowcasting. In NIPS, pages 802 – 810, 2015