Image Super-Resolution Using Deep Convolutional Networks. Dong, C., Loy, Change, C., He, K., & Tang, X. arXiv:1501.00092 [cs], December, 2014. 00000 arXiv: 1501.00092Paper abstract bibtex We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.
@article{ dong_image_2014,
title = {Image {Super}-{Resolution} {Using} {Deep} {Convolutional} {Networks}},
url = {http://arxiv.org/abs/1501.00092},
abstract = {We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.},
urldate = {2015-01-14TZ},
journal = {arXiv:1501.00092 [cs]},
author = {Dong, Chao and Loy, Chen Change and He, Kaiming and Tang, Xiaoou},
month = {December},
year = {2014},
note = {00000
arXiv: 1501.00092},
keywords = {deep learning, reading}
}
Downloads: 0
{"_id":"cPbpcaEirhkjmxK9f","bibbaseid":"dong-loy-change-he-tang-imagesuperresolutionusingdeepconvolutionalnetworks-2014","downloads":0,"creationDate":"2015-09-03T07:18:49.087Z","title":"Image Super-Resolution Using Deep Convolutional Networks","author_short":["Dong, C.","Loy","Change, C.","He, K.","Tang, X."],"year":2014,"bibtype":"article","biburl":"http://bibbase.org/zotero/fred.qi","bibdata":{"abstract":"We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.","author":["Dong, Chao","Loy","Change, Chen","He, Kaiming","Tang, Xiaoou"],"author_short":["Dong, C.","Loy","Change, C.","He, K.","Tang, X."],"bibtex":"@article{ dong_image_2014,\n title = {Image {Super}-{Resolution} {Using} {Deep} {Convolutional} {Networks}},\n url = {http://arxiv.org/abs/1501.00092},\n abstract = {We propose a deep learning method for single image super-resolution (SR). Our method directly learns an end-to-end mapping between the low/high-resolution images. The mapping is represented as a deep convolutional neural network (CNN) that takes the low-resolution image as the input and outputs the high-resolution one. We further show that traditional sparse-coding-based SR methods can also be viewed as a deep convolutional network. But unlike traditional methods that handle each component separately, our method jointly optimizes all layers. Our deep CNN has a lightweight structure, yet demonstrates state-of-the-art restoration quality, and achieves fast speed for practical on-line usage. We explore different network structures and parameter settings to achieve trade-offs between performance and speed. Moreover, we extend our network to cope with three color channels simultaneously, and show better overall reconstruction quality.},\n urldate = {2015-01-14TZ},\n journal = {arXiv:1501.00092 [cs]},\n author = {Dong, Chao and Loy, Chen Change and He, Kaiming and Tang, Xiaoou},\n month = {December},\n year = {2014},\n note = {00000 \narXiv: 1501.00092},\n keywords = {deep learning, reading}\n}","bibtype":"article","id":"dong_image_2014","journal":"arXiv:1501.00092 [cs]","key":"dong_image_2014","keywords":"deep learning, reading","month":"December","note":"00000 arXiv: 1501.00092","title":"Image Super-Resolution Using Deep Convolutional Networks","type":"article","url":"http://arxiv.org/abs/1501.00092","urldate":"2015-01-14TZ","year":"2014","bibbaseid":"dong-loy-change-he-tang-imagesuperresolutionusingdeepconvolutionalnetworks-2014","role":"author","urls":{"Paper":"http://arxiv.org/abs/1501.00092"},"keyword":["deep learning","reading"],"downloads":0},"search_terms":["image","super","resolution","using","deep","convolutional","networks","dong","loy","change","he","tang"],"keywords":["deep learning","reading"],"authorIDs":[],"dataSources":["y8en6y5RHukfeLuPH"]}