BEGIN:VCALENDAR

VERSION:2.0

PRODID:-//wp-events-plugin.com//6.6.4.4//EN

TZID:Asia/Jerusalem

X-WR-TIMEZONE:Asia/Jerusalem
BEGIN:VEVENT

UID:0-428@aerospace.technion.ac.il

DTSTART;TZID=Asia/Jerusalem:20170802T163000

DTEND;TZID=Asia/Jerusalem:20170802T173000

DTSTAMP:20230530T180843Z

URL:https://aerospace.technion.ac.il/events/bundle-adjustment-with-feature
 -scale-constraint-for-enhanced-estimation-accuracy/

SUMMARY:Bundle Adjustment with Feature Scale Constraint for Enhanced Estima
 tion Accuracy
DESCRIPTION:Lecturer:Vladimir Ovechkin\n Faculty:Technion Autonomous System
 s Program (TASP)\,\n Institute:Technion – Israel Institute of Technology
 \n Location:Classroom 165\, ground floor\, Library\, Aerospace Eng.\n Zoom
 : \n Abstract: \n Details: \n Accurate pose estimation and 3D reconstructi
 on are important in a variety of applications such as autonomous navigatio
 n or mapping in uncertain or unknown environments. Bundle adjustment (BA) 
 and simultaneous localization and mapping (SLAM) are commonly used approac
 hes to address these and other related problems. Given a sequence of image
 s\, BA is the problem of simultaneously inferring the camera poses and the
  observed 3D landmarks. BA is typically solved by minimizing the re-projec
 tion error between image observations (image features) and their predictio
 n obtained by projecting the corresponding landmarks onto the camera frame
 . This optimization is typically realized using non-linear least-squares (
 NLS) approaches. Different techniques exist for detecting image features\
 , including the well-known SIFT features. Yet\, state of the art BA and vi
 sual SLAM approaches formulate the constraints in the NLS optimization uti
 lizing only image feature coordinates.\nIn this work we propose to incorpo
 rate within BA a new type of constraints that use feature scale informatio
 n that is readily available from a typical image feature detector (e.g. SI
 FT\, SURF). While feature scales (and orientation) play an important role 
 in image matching\, they have not been utilized thus far for estimation p
 urposes in BA framework. Our approach exploits feature scale information a
 nd uses it to enhance the accuracy of bundle adjustment\, especially along
  the optical axis of the camera in a monocular setup. Specifically\, we f
 ormulate constraints between the measured and predicted feature scale\, wh
 ere the latter depends on the distance from the camera and the correspondi
 ng 3D point\, and optimize the system variables to minimize the residual e
 rror in these constraints in addition to minimizing the standard re-proje
 ction error. We study our approach both in synthetic environments and real
 -image ground and aerial datasets (KITTI and Kagaru)\, demonstrating signi
 ficant improvement in positioning error.
CATEGORIES:Seminars
LOCATION:Classroom 165\, ground floor\, Library\, Aerospace Eng.

END:VEVENT

BEGIN:VTIMEZONE

TZID:Asia/Jerusalem

X-LIC-LOCATION:Asia/Jerusalem

BEGIN:DAYLIGHT

DTSTART:20170324T030000

TZOFFSETFROM:+0200

TZOFFSETTO:+0300

TZNAME:IDT

END:DAYLIGHT

END:VTIMEZONE
END:VCALENDAR