eduzhai > Applied Sciences > Engineering >

Mapping Low-Resolution Images To Multiple High-Resolution Images Using Non-Adversarial Mapping

  • Save

... pages left unread,continue reading

Document pages: 17 pages

Abstract: Several methods have recently been proposed for the Single ImageSuper-Resolution (SISR) problem. The current methods assume that a singlelow-resolution image can only yield a single high-resolution image. Inaddition, all of these methods use low-resolution images that were artificiallygenerated through simple bilinear down-sampling. We argue that, first andforemost, the problem of SISR is an one-to-many mapping problem between thelow-resolution and all possible candidate high-resolution images and we addressthe challenging task of learning how to realistically degrade and down-samplehigh-resolution images. To circumvent this problem, we propose SR-NAM whichutilizes the Non-Adversarial Mapping (NAM) technique. Furthermore, we propose adegradation model that learns how to transform high-resolution images tolow-resolution images that resemble realistically taken low-resolution photos.Finally, some qualitative results for the proposed method along with theweaknesses of SR-NAM are included.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×