• DocumentCode
    2494134
  • Title

    Urban models texturing from un-calibrated photographs

  • Author

    Moslah, O. ; Klee, M. ; Grolleau, A. ; Guitteny, V. ; Couvet, S. ; Philipp-Foliguet, S.

  • Author_Institution
    Thales Security Solutions & Services, Cergy-Pontoise
  • fYear
    2008
  • fDate
    26-28 Nov. 2008
  • Firstpage
    1
  • Lastpage
    6
  • Abstract
    We present a system to automatically fit a CAD urban model to a structure from motion sparse reconstruction. Fitting the model to the sparse reconstruction is defined and solved as an absolute orientation problem. The novelty of our approach is the use of facade correspondences to compute the absolute orientation. We also propose a solution to automatically establish facade correspondences using a pseudo-RANSAC procedure. Our system consists of four parts: a robust structure from motion recovery framework able to register accurately a set of digital photographs, a method to recover the principal walls of a scene from a 3d sparse reconstruction using a robust RANSAC scheme, a method to fit automatically a 3d model to a set of 3d wall planes, and finally, a user interface allowing scene visualisation and photorealistic rendering of urban 3d models from a set of calibrated photographs.
  • Keywords
    architectural CAD; image motion analysis; image reconstruction; photography; rendering (computer graphics); town and country planning; CAD urban model; absolute orientation problem; motion recovery; motion sparse reconstruction; photorealistic rendering; pseudo-RANSAC procedure; scene visualisation; uncalibrated photographs; urban models texturing; user interface; Cameras; Context modeling; Design automation; Image reconstruction; Layout; Rendering (computer graphics); Robustness; Security; Urban planning; Visualization; 3d plane fitting; absolute orientation; structure from motion;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Image and Vision Computing New Zealand, 2008. IVCNZ 2008. 23rd International Conference
  • Conference_Location
    Christchurch
  • Print_ISBN
    978-1-4244-3780-1
  • Electronic_ISBN
    978-1-4244-2583-9
  • Type

    conf

  • DOI
    10.1109/IVCNZ.2008.4762069
  • Filename
    4762069