Skip to Main content Skip to Navigation
Conference papers

Une nouvelle méthode de classification en grande dimension pour la reconnaissance de formes

Charles Bouveyron 1, 2 Stéphane Girard 2 Cordelia Schmid 1, * 
* Corresponding author
1 LEAR - Learning and recognition in vision
GRAVIR - IMAG - Laboratoire d'informatique GRAphique, VIsion et Robotique de Grenoble, Inria Grenoble - Rhône-Alpes, CNRS - Centre National de la Recherche Scientifique : FR71
Abstract : We propose a new Gaussian model to classify high-dimensional data in both supervised and unsupervised frameworks. Our approach is based on the assumption that high-dimensional data live in low-dimensional subspaces. Our model therefore finds the specific subspace and the intrinsic dimension of each class to correctly fit the data. In addition, our approach regularizes the class conditional covariance matrices by assuming that classes are spherical both in their eigenspace and in its supplementary. We thus obtain a robust clustering method for high-dimensional data. Our approach is then applied to recognize object in real images and its performances are compared to classical methods.
Document type :
Conference papers
Complete list of metadata
Contributor : THOTH Team Connect in order to contact the contributor
Submitted on : Monday, December 20, 2010 - 9:08:45 AM
Last modification on : Thursday, January 20, 2022 - 5:30:12 PM


  • HAL Id : inria-00548515, version 1



Charles Bouveyron, Stéphane Girard, Cordelia Schmid. Une nouvelle méthode de classification en grande dimension pour la reconnaissance de formes. 20ème colloque GRETSI sur le traitement du signal et des images, Sep 2005, Louvain-la-Neuve, Belgique. pp.365-368. ⟨inria-00548515⟩



Record views