Complementing SRCNN by Transformed Self-Exemplars

Publication: Research - peer-reviewArticle in proceeding

Abstract

Super-resolution algorithms are used to improve the qualityand resolution of low-resolution images. These algorithms can be dividedinto two classes of hallucination- and reconstruction-based ones. Theimprovement factors of these algorithms are limited, however, previousresearch [10], [9] has shown that combining super-resolution algorithmsfrom these two different groups can push the improvement factor further.We have shown in this paper that combining super-resolution algorithmsof the same class can also push the improvement factor up. For thispurpose, we have combined two hallucination based algorithms, namelythe one found in Single Image Super-Resolution from Transformed Self-Exemplars [7] and the Super-Resolution Convolutional Neural Networkfrom [4]. The combination of these two, through an alpha-blending, hasresulted in a system that outperforms state-of-the-art super-resolutionalgorithms on public benchmark datasets.
Close

Details

Super-resolution algorithms are used to improve the qualityand resolution of low-resolution images. These algorithms can be dividedinto two classes of hallucination- and reconstruction-based ones. Theimprovement factors of these algorithms are limited, however, previousresearch [10], [9] has shown that combining super-resolution algorithmsfrom these two different groups can push the improvement factor further.We have shown in this paper that combining super-resolution algorithmsof the same class can also push the improvement factor up. For thispurpose, we have combined two hallucination based algorithms, namelythe one found in Single Image Super-Resolution from Transformed Self-Exemplars [7] and the Super-Resolution Convolutional Neural Networkfrom [4]. The combination of these two, through an alpha-blending, hasresulted in a system that outperforms state-of-the-art super-resolutionalgorithms on public benchmark datasets.
Original languageEnglish
Title of host publicationInternational Conference on Pattern Recognition (ICPR) : workshop on Face and Facial Expression Recognition
Publication date28 Mar 2017
ISBN (print)978-3-319-56686-3
ISBN (electronic)978-3-319-56687-0
StatePublished - 28 Mar 2017
Event23rd international conference on pattern recognition (ICPR 2016) - Cancun, Mexico

Conference

Conference23rd international conference on pattern recognition (ICPR 2016)
LandMexico
ByCancun
Periode04/12/201608/12/2016
ID: 243277400