Journal or Publication Title:
Date of publication:
Volume:
ISSN:
Additional Information:
Սհոյան Միսակ Տ., Сгоян Мисак Т.
Title:
Other title:
Պատկերից շարժման հետևանքով առաջացած շաղվածության հեռացում և պատկերի կետայնության բարձրացում՝ օգտագործելով շերտային ուշադրության բազմամասշտաբ մոդուլները ; Удаление размытости вызваннойд вижением и увеличение разрешения одного изображения с использованием много масштабных модулей внимания канала
Creator:
Subject:
Mathematical cybernetics ; Computer science
Uncontrolled Keywords:
Keywords:Motion deblurring ; super-resolution ; channel attention
Abstract:
During the last decade, deep convolutional neural networks have significantly advanced the single image super-resolution techniques reconstructingrealistic textural and spatial details. In classical image super-resolution problems, it is assumed that the low-resolution image has a certaindownsampling degradation. However, complicated image degradations are inevitable in real-world scenarios,and motion blur is a common type of image degradation due to camera or scene motion during the image capturing process. This work proposes a fully convolutional neural networkto reconstructhigh-resolution sharp images from thegiven motion blurry low-resolution images. The deblurring subnetworkisbased on multi-stage progressive architecture, while the super-resolution subnetworkis designed using the multi-scale channel attention modules. A simple andeffective training strategy is employedwhere a pre-trained frozen deblurring module is used to train the super-resolutionmodule.The deblurring module is unfrozen inthe last training phase. Experiments show that,unlike the other methods, the proposed method reconstructs relatively small structures and textural details while successfully removing the complex motion blur. The implementation code and the pre-trained model are publicly available at https://github.com/misakshoyan/joint-motion-deblur-and-sr.