BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:US/Eastern
BEGIN:DAYLIGHT
DTSTART:20180311T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:EDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20171105T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:EST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20180503T000000Z
UID:7A1BAA02-EC8C-4486-8A14-587962A40D10
DTSTART;TZID=US/Eastern:20180302T123000
DTEND;TZID=US/Eastern:20180302T133000
DESCRIPTION:Image processing is fundamental to research in areas such as ne
 utron imaging\, computer vision\, medical imaging\, and remote sensing. Ho
 wever\, while many core image processing concepts are transferable across 
 domains\, appropriate implementation may require unique modifications to a
 dapt the algorithms to the domain phenomenology.\n\nIn this talk I will fo
 cus on how feature-based image registration methods differ for various app
 lications\, and how they can be adapted for optimal performance. I will de
 scribe the concept of control point or “keypoint” matching from a gene
 ral perspective. I will first discuss its implementation in the computer v
 ision domain. I will then show the limitations of applying this implementa
 tion directly to applications in remote sensing. Based on these limitation
 s\, a second implementation will then be proposed from the remote sensing 
 point of view\, and conclusions will be drawn related to algorithm transfe
 rability. This talk will build an intuition for adapting generic image pro
 cessing techniques to specific domains of interest.\n\nSpeaker(s): Sophie 
 Voisin\, \n\nRoom: 1275\, Bldg: Center for Imaging Science Bld 76\, Roches
 ter Institute of Technology\, 54 Lomb Memorial Drive\, Rochester\, New Yor
 k\, United States\, 14623
LOCATION:Room: 1275\, Bldg: Center for Imaging Science Bld 76\, Rochester I
 nstitute of Technology\, 54 Lomb Memorial Drive\, Rochester\, New York\, U
 nited States\, 14623
ORGANIZER:emmett@cis.rit.edu
SEQUENCE:7
SUMMARY:image processing: adapting keypoint matching across domains
URL;VALUE=URI:https://events.vtools.ieee.org/m/169003
X-ALT-DESC:Description: <br /><p>&nbsp\;&nbsp\; Image processing is fundame
 ntal to research in areas such as neutron imaging\, computer vision\, medi
 cal imaging\, and remote sensing. However\, while many core image processi
 ng concepts are transferable across&nbsp\; domains\, appropriate implement
 ation may require unique modifications to adapt the algorithms to the doma
 in phenomenology.</p>\n<p>&nbsp\;&nbsp\;&nbsp\; In this talk I will focus 
 on how <strong>feature-based image registration </strong>methods differ fo
 r various applications\, and how they can be adapted for optimal performan
 ce. I will describe the concept of control point or &ldquo\;keypoint&rdquo
 \; matching from a general perspective. I will first discuss its implement
 ation in the <strong>computer vision </strong>domain. I will then show the
  limitations of applying this implementation directly to applications in <
 strong>remote sensing</strong>. Based on these limitations\, a second impl
 ementation will then be proposed from the remote sensing point of view\, a
 nd conclusions will be drawn related to algorithm transferability. This ta
 lk will build an intuition for adapting generic image processing technique
 s to specific domains of interest.</p>
END:VEVENT
END:VCALENDAR