|Title: Using Projective Differential Geometry to Model Deformable Surfaces and their Images||Posted: September 10, 2014|
|Company/Institution: Universite d'Auvergne|
|Location: Clermont-Ferrand, France|
|Department: ISIT - Image Science for Interventional Techniques|
Description: This postdoc will take place in the context of a 5 year Computer Vision project named FLEXABLE, focusing on the 3D reconstruction of deformable objects from multiple images. The key problem is to estimate the 3D shape of a deforming surface from its 2D projections. This problem is known as Non-Rigid Shape-from-Motion (NRSfM) in Computer Vision and is largely open. One of the current challenges is modeling. The surface of many objects may be modeled as a Riemannian manifold. However, as images are essentially created by perspective projection, the Riemannian structure is then lost in the 2D projection, and is replaced by a locally projective structure. The goal of this postdoc will be to study how the framework of projective manifolds may be used to model deformable surfaces embedded in 3D, their 2D projections, and the relationship existing between their 2D projections.
The postdoc is funded by the EU through the excellence ERC grant FLEXABLE. It will take place in the ALCoV group under the supervision of Prof Adrien Bartoli (http://isit.u-clermont1.fr/~ab) and Dr Daniel Pizarro. ALCoV is a young and extremely dynamical research group, which has been pioneering the field of 3D computer vision in deformable environments and optical 3D reconstruction in laparoscopy. Publications at an excellent level will be encouraged.
The ideal candidate has a PhD in the field of mathematics or computer science.
Application Instructions: Send your CV including the name of at least two references, one of your publications, and any other relevant documents by email to Adrien Bartoli. The recruitment will be done as soon as an excellent candidate is found. The salary is competitive, and the duration may be up to 3.5 years.