Vous êtes ici : Accueil » Kiosque » Annonce

Identification

Identifiant: 
Mot de passe : 

Mot de passe oublié ?
Détails d'identification oubliés ?

Annonce

30 novembre 2017

Stage M2 at Inria Rennes: "mesh prediction and coding for interactive 3D object display system"


Catégorie : Stagiaire


Master project in the team SIROCCO (Inria Rennes)

Title : mesh prediction and coding for interactive 3D object display system

Summary: In the context of free viewpoint television, we will study techniques to predict and encode 3D meshes, in a suitable way for interactive navigation.

Contact: thomas.maugey@inria.fr

 

Introduction

Free viewpoint visualization is a system for multimedia content transmission, in which the user can observe a scene from the angle of view he desires. This type of application is really attractive and more and more studied since it brings a sensation of immersion to the users, while having a control on the observed data. In order to be efficient, this system necessitates an interactive coding scheme that only transmits to the user the data which are still unknown and that appears in its field of view during the navigation. In other words, only what is required is transmitted. In order to send the minimum amount of information at every viewpoint change, a classical approach relies on a prediction tool at the client side: from the data in memory when watching viewpoint i, the decoder predicts what will be seen at the new viewpoint i+1 , and only the prediction error is sent.

Project context

As a preliminary to this master project, an interactive visualization system has been developed for 3D object. This system deals with the interactivity as detailed above:

  1. We only display the part of the 3D object that is visible from the user point of view
  2. If the user moves, the newly visible parts are loaded.

This system also manages a simple version of the prediction: the new part of the mesh is estimated based on the one previously available in memory.

Objectives

The objectives of the Master project are:

[1] A. Roumy, T. Maugey, Universal lossless coding with random user access: the cost of interactivity, IEEE ICIP, Quebec, Canada, Sep. 2015.

[2] N. Mahmoudian Bidgoli, T. Maugey, A. Roumy Correlation Model Selection for interactive video communication, ICIP, Beijing, China, Sept. 2017

This project will be located in the Inria Research Laboratory in Rennes (France), and supervised by Thomas Maugey.

For more information and for applications, please Thomas Maugey: thomas.maugey@inria.fr.

 

Dans cette rubrique

(c) GdR 720 ISIS - CNRS - 2011-2018.