Lightweight  Network  Based  on  Interleaved  Group Convolution for Image Super-Resolution

Jiexin Zhang; HanWang; Jiaxin Luo; Zheng Zhang

doi:10.18686/ahe.v7i13.8462

Lightweight Network Based on Interleaved Group Convolution for Image Super-Resolution

Jiexin Zhang, HanWang , Jiaxin Luo, Zheng Zhang

Abstract

Deep learning has been successfully applied to single im- age super-resolution problems due to its high data fi tting abil-ity. However, the increasing depth and complexity of the network has brought about the disappearance of information, data volume and computational redundancy, and is not suit- able for small devices. To solve these problems, we propose a new lightweight network model based on interleaved group convolution for single image super-resolution reconstruction. The core idea of this algorithm is to broaden the network structure by means of group convolution, enhance the sparsity of the convolution kernel, and achieve the purpose of reduc- ing the amount of calculation and the amount of parameters. After a lot of experimental evaluation,we prove that our al- gorithm can achieve better results with a smaller number of parameters.

Keywords

Super-Resolution; Deep-learning; Lightwe- ight model; Residual Learning; Interleaved group convolution

Full Text:

PDF

Included Database

References

[1] Francesc Ara `ndiga, “A nonlinear algorithm for mono- tone piecewise bicubic interpolation,” Applied Mathe- matics and Computation, vol. 272, pp. 100–113, 2016.

[2] Xin Yang, Yan Zhang, Dake Zhou, and Ruigang Yang, “An improved iterative back projection algorithm based on ringing artifacts suppression,” Neurocomputing, vol. 162, pp. 171–179, 2015.

[3] Chao Dong, Chen Change Loy, Kaiming He, and Xi- aoou Tang, “Learning a deep convolutional network for image super-resolution,” in Computer Vision - ECCV 2014 - 13th European Conference, 2014, pp. 184–199.

[4] Chao Dong, Chen Change Loy, and Xiaoou Tang, “Ac- celerating the super-resolution convolutional neural net- work,” in Computer Vision - ECCV 2016 - 14th Euro- pean Conference, Amsterdam, 2016, pp. 391–407.

[5] Ting Zhang, Guo-Jun Qi, Bin Xiao, and Jingdong Wang, “Interleaved group convolutions for deep neural net- works,” CoRR, vol.abs/1707.02725, 2017.

[6] Tong Tong, Gen Li, Xiejie Liu, and Qinquan Gao, “Im- age super-resolution using dense skip connections,” in IEEE International Conference on Computer Vision, ICCV, 2017, pp. 4809–4817.

[7] Zhang Y Liu SG Guo M. Peng YL, Zhang L, “Deep deconvolution neural network for image super- resolution.(in chinese).,”Journal of Software, pp. 29(4): 926–934, 2018.

[8] Jiwon Kim, Jung Kwon Lee, and Kyoung Mu Lee, “Ac- curate image super-resolution using very deep convolu- tional networks,”in 2016 IEEE Conference on Com- puter Vision and Pattern Recognition,CVPR, 2016, pp. 1646–1654.

[9] Radu Timofte, Eirikur Agustsson, and et al, “NTIRE 2017 challenge on single image super-resolution: Meth- ods and results,” in 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops, CVPR, 2017, pp. 1110–1121.

[10] Marco Bevilacqua, Aline Roumy, Christine Guillemot, and Marie-Line Alberi-Morel, “Low-complexity single- image super-resolution based on nonnegative neighbor embedding,” in British Machine Vision Conference, BMVC, 2012, pp. 1–10.

[11] Roman Zeyde, Michael Elad, and Matan Protter, “On single image scale-up using sparse-representations,” in Curves and Surfaces- 7th International Conference, 2010, pp. 711–730.

[12] David R. Martin, Charless C. Fowlkes, Doron Tal, and Jitendra Malik, “A database of human segmented nat- ural images and its application to evaluating segmenta- tion algorithms and measuring ecological statistics,” in 2001 IEEE International Conference on Computer Vi- sion, ICCV, 2001, pp. 416–425.

[13] Jia-Bin Huang, Abhishek Singh, and Narendra Ahuja, “Single image super-resolution from transformed self-exemplars,” in IEEE Conference on Computer Vision and Pattern Recognition, CVPR, 2015, pp. 5197–5206.

[14] Kevin Jarrett, Koray Kavukcuoglu, Marc’Aurelio Ran- zato, and Yann LeCun, “What is the best multi-stage ar- chitecture for object recognition?,” in IEEE 12th Inter- national Conference on Computer Vision, ICCV, 2009, pp. 2146–2153.

[15] Diederik P. Kingma and Jimmy Ba, “Adam: A method for stochastic optimization,” CoRR, vol. abs/1412.6980, 2014.

[16] Wei-Sheng Lai, Jia-Bin Huang, Narendra Ahuja, and Ming-Hsuan Yang, “Deep laplacian pyramid networks for fast and accurate super-resolution,” in 2017 IEEE Conference on Computer Vision and Pattern Recogni- tion,CVPR, 2017, pp. 5835–5843.

DOI: http://dx.doi.org/10.18686/ahe.v7i13.8462

Refbacks

There are currently no refbacks.