BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.bham.ac.uk//v3//EN
BEGIN:VEVENT
CATEGORIES:Medical Imaging Research Seminars
SUMMARY:Regularisations in image processing and NIRFAST. -
Wenqi Lu\, Doctoral researcher\, Medical Imaging
and Image Interpretation Group \,School of Compute
r Science\, University of Birmingham
DTSTART:20161011T100000Z
DTEND:20161011T110000Z
UID:TALK2235AT
URL:/talk/index/2235
DESCRIPTION:Inverse problem in NIRFAST is a highly under-deter
mined ill-posed mathematical problem. It aims to r
econstruct certain optical properties on each fini
te element node from a small amount of boundary me
asurements. It is hard to determine a unique solut
ion for this inverse problem. Utilising regularisa
tions is therefore imperative to alleviate this di
fficulty and help find a desirable solution for su
ch ill-posed mathematical problem. This talk will
go through my MSc research on regularisations in i
mage processing and show some detailed comparison
on some representative regularisations. Discretisa
tion process based on discrete finite difference m
ethod (FDM) and numerical implementation of altern
ating direction method of multipliers (ADMM) algor
ithm are also briefly introduced. Whether or not t
hese regularisations can be applied to the inverse
problem in NIRFAST and the possible solutions wil
l be discussed finally.
LOCATION:Sloman Lounge\, Computer Science Building (Y9 in
Edgbaston Campus)
CONTACT:Matthaios Doulgerakis
END:VEVENT
END:VCALENDAR