Home Grants
 
Home Grants

3D modeling and location

Research Area: Computer Vision
Status: In progress  
Faculty: Francesca Odone Participants: Alessandro Verri, Luca Zini, Nicoletta Noceti
 
Description:

The starting point is our long-standing experience in the field of 3D computer vision including motion analysis, 3D reconstruction, and object recognition.

We are now exploring the use of prior information on the geometry of the viewed scene to the purpose of detecting accurately location and pose of objects of interest. A main application setting we are considering is planetary Entry Descent and Landing (EDL). Within this  challenging research activity we focus in particular on:

  • relative pose estimation, where we deal with the problem of estimating the relative pose of a spacecraft from image points often in quasi-degenerate configurations (in particular lying on planar surfaces)
  • absolute pose estimation, in this case the final goal is to precisely locate the spacecraft with respect to the planet geography. Here we address in particular the problem of detecting landmarks on the surface.
  • computer vision methods for safe landing, here in particular we address the problem of surface slope estimation and rock detection.

We are also looking at the use of computational algebraic techniques for studying the geometric structure of the viewed scene.


Ongoing collaborations

  • Thales Alenia Space spa (Torino, IT)

 

 


 

References

  • Odone, Verri, Zini, Lanza and Mercer "Relative pose estimation for planetary Entry Descent and Landing", ACCV - Workshop on "Computer Vision for Vehicle Tecnologies: from Earth to Mars", 2010
  • Delponte, E. et al. "SVD-matching using SIFT features". Graphical Models 68 (2006): 415-431.
  • Delponte, E. et al. "Spatio-temporal constraints for matching view-based descriptions of 3D object". WIAMIS, 2007