Deutsch
 
Hilfe Datenschutzhinweis Impressum
  DetailsucheBrowse

Datensatz

DATENSATZ AKTIONENEXPORT
Zur Ablage hinzufügen
  Lokale TagsFreigabegeschichteDetailsÜbersicht

Freigegeben
Konferenzbeitrag

Exploiting Photogrammetric Targets for Industrial AR

MPG-Autoren
Es sind keine MPG-Autoren in der Publikation vorhanden
Externe Ressourcen
Es sind keine externen Ressourcen hinterlegt
Volltexte (beschränkter Zugriff)
Für Ihren IP-Bereich sind aktuell keine Volltexte freigegeben.
Volltexte (frei zugänglich)
Es sind keine frei zugänglichen Volltexte in PuRe verfügbar
Ergänzendes Material (frei zugänglich)
Es sind keine frei zugänglichen Ergänzenden Materialien verfügbar
Zitation

Naik, H., Oyamada, Y., Keitler, P., & Navab, N. (2015). Exploiting Photogrammetric Targets for Industrial AR. In C. {Sandor, R. Lindeman, W. Mayol-Cuevas, N. Sakata, R. Newcombe, & V. Teichrieb (Eds.), 2015 IEEE INTERNATIONAL SYMPOSIUM ON MIXED AND AUGMENTED REALITY (pp. 144-147).


Zitierlink: https://hdl.handle.net/21.11116/0000-0007-5B88-A
Zusammenfassung
In this work, we encourage the idea of using Photogrammetric targets for
object tracking in Industrial Augmented Reality (IAR). Photogrammetric
targets, especially uncoded circular targets, are widely used in the
industry to perform 3D surface measurements. Therefore, an AR solution
based on the uncoded circular targets can improve the work flow
integration by reusing existing targets and saving time. These circular
targets do not have coded patterns to establish unique 2D-3D
correspondences between the targets on the model and their image
projections. We solve this particular problem of 2D-3D correspondence of
non-coplanar circular targets from a single image. We introduce a Conic
pair descriptor, which computes the Eucledian invariants from circular
targets in the model space and in the image space. A three stage method
is used to compare the descriptors and compute the correspondences with
up to 100% precision and 89% recall rates. We are able to achieve
tracking performance of 3 FPS (2560x1920 pix) to 8 FPS (640x480 pix)
depending on the camera resolution and the targets present in the scene.