Visual Augmentation of Complex Environments

Magrit is an INRIA research group dedicated to computer vision.

It is part of the INRIA Nancy - Grand Est research center and LORIA laboratory.


The Magrit team carries out research into the field of computer vision, with a focus on augmented reality applications. Augmented reality is a relatively new field. It aims to augment the user's perception by adding in his field of view elements that improve his comprehension of his environment. Applications of this concept are plentiful and concern medical gesture assistance, learning and maintenance systems, cultural heritage, audiovisual... In order to integrate information at the right place in the user's field of view, whatever his motion, the observer's viewpoint has to be computed at every instant. Moreover, reconstructing, even partially, the observed environment is necessary to manage occlusions between the added objects and the scene or to take light interreflexions into account. Only elementary commercial applications of this concept exist to date. These applications are only effective in environments with limited user action which are furthermore often instrumented (landmarks). Many challenges remain to be tackled so as to address applications in complex environments. The Magrit team research aims to propose robust solutions to the two main issues faced in augmented reality: viewpoint computation and reconstruction of the scene elements necessary to set up the application. More precisely, our research themes deal with:

  • sequential and real time viewpoint computation, in structured and unstructured environments.
  • management of long time running applications: one of the recurrent issues is viewpoint drift along time. Our objective is to improve the accuracy and repeatability of viewpoint computation for long time running applications.
  • 3D modeling for augmented reality applications: the problem is fundamental to handle interactions between the real and virtual worlds.
  • information fusion: in any augmented reality application, careful thoughts must be taken about the saliency of the added information and about the best way to integrate it in the real image. This issue is a key element in the framework of our medical applications, where validation of the augmented images is crucial.


Marie-Odile Berger
tél : (+33) (0)3 54 95 85 01
CNRS INRIA Université de Lorraine LORIA

copyright INRIA / Photos C. Lebedinsky