BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Cambridge Analysts' Knowledge Exchange
SUMMARY:Reverse Engineering a PDE from an Image Inpainting
Algorithm - Rob Hocking\, CCA
DTSTART;TZID=Europe/London:20150610T160000
DTEND;TZID=Europe/London:20150610T170000
UID:TALK59281AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/59281
DESCRIPTION:A common practice in designing image inpainting me
thods is to model the image as a continuum\, where
the completed image may be expressed in terms of
the solution to a suitable PDE. One then discreti
zes this PDE to obtain an algorithm that may be ap
plied to real images\, which are discrete. Howeve
r\, what about working in reverse? Given a discre
te algorithm\, can we find a PDE such that images
inpainted via this algorithm converge to solutions
to said PDE in the continuum limit? If so\, can
this help us to understand our method or improve i
t?\n\nIn this talk\, I will share my experience wo
rking on inpainting algorithms for industry\, wher
e people may not have heard of PDEs and methods ar
e designed based primarily on intuition. I will s
how how these methods may nonetheless be understoo
d in terms of PDEs\, and furthermore that this und
erstanding can be utilized to create improved meth
ods. Finally\, I will argue that this may give yo
u a better algorithm in the end than if you simply
started from a desired PDE and and discretized it
in a routine manner.
LOCATION:MR11\, Centre for Mathematical Sciences
CONTACT:Dominic Dold
END:VEVENT
END:VCALENDAR