Suivi d'objets en imagerie aérienne

Aude Jacquot 1
1 PERCEPTION - Interpretation and Modelling of Images and Videos
GRAVIR - IMAG - Graphisme, Vision et Robotique, Inria Grenoble - Rhône-Alpes, CNRS - Centre National de la Recherche Scientifique : FR71
Abstract : This thesis results from a CIFRE collaboration with Thales Optronique, a compnay which develops airborne systems for "laser guided armaments". The operating mode used by these systems requires on behalf of the pilot an important workload. The objective of this thesis is to reduce this responsibility, by making the current trackers more robust to the three dimensional effects, in particular in the presence of reliefs.

The main theme of this thesis is the tracking of objects from aerial images. We want to extract some three dimensional information from video sequences in order to improve the existing algorithms. For that purpose, we place ourselves in a Bayesian framework and formulate the tracking problem by means of a particle filter. We set up three algorithms: The first one is based on geometrical models (that can be 2D or 3D), that we combine with a particle filter. We add a supplementary stage to the classical particle filter, allowing us to change model when the algorithm judges it necessary. And the last one combines both precedents; the integration of histograms and contour information in a particle filter allows not only to have a more robust tracker, but also to take into account the 3D real information of the observed scene.

An evaluation protocol has been set up to estimate the performances of these algorithms. Results illustrate the performances of these algorithms.
Document type :
Theses
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00379479
Contributor : Peter Sturm <>
Submitted on : Tuesday, April 28, 2009 - 4:55:07 PM
Last modification on : Wednesday, July 11, 2018 - 1:15:58 AM
Long-term archiving on : Thursday, June 10, 2010 - 6:51:13 PM

Identifiers

  • HAL Id : tel-00379479, version 1

Collections

INRIA | UGA | IMAG

Citation

Aude Jacquot. Suivi d'objets en imagerie aérienne. Interface homme-machine [cs.HC]. Institut National Polytechnique de Grenoble - INPG, 2006. Français. ⟨tel-00379479⟩

Share

Metrics

Record views

509

Files downloads

313