BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Australia/Brisbane
X-LIC-LOCATION:Australia/Brisbane
BEGIN:STANDARD
DTSTART:19920301T030000
TZOFFSETFROM:+1100
TZOFFSETTO:+1000
TZNAME:AEST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20191203T080450Z
LOCATION:Plaza Meeting Room P2
DTSTART;TZID=Australia/Brisbane:20191118T114200
DTEND;TZID=Australia/Brisbane:20191118T120300
UID:siggraphasia_SIGGRAPH Asia 2019_sess104_papers_412@linklings.com
SUMMARY:Blind image super resolution with spatially variant degradations
DESCRIPTION:Technical Papers\n\nBlind image super resolution with spatiall
 y variant degradations\n\nCornillère, Djelouah, Wang, Schroers, Sorkine-Ho
 rnung\n\nExisting deep learning approaches to single image super-resolutio
 n have achieved impressive results but mostly assume a setting with fixed 
 pairs of high resolution and low resolution images. However, to robustly a
 ddress realistic upscaling scenarios where the relation between high resol
 ution and low resolution images is unknown, blind image super-resolution i
 s required. To this end, we propose a solution that relies on three compon
 ents: First, we use a degradation aware SR network to synthesize the HR im
 age given a low resolution image and the corresponding blur kernel. Second
 , we train a kernel discriminator to analyze the generated high resolution
  image in order to predict errors present due to providing an incorrect bl
 ur kernel to the generator. Finally, we present an optimization procedure 
 that is able to recover both the degradation kernel and the high resolutio
 n image by minimizing the error predicted by our kernel discriminator. We 
 also show how to extend our approach to spatially variant degradations tha
 t typically arise in visual effects pipelines when compositing content fro
 m different sources and how to enable both local and global user interacti
 on in the upscaling process.\n\nRegistration Category: Platinum Pass, Full
  Conference Pass, Full Conference One-Day Pass
URL:https://sa2019.conference-program.com/presentation?id=papers_412&sess=
 sess104
END:VEVENT
END:VCALENDAR