National audienceA kernel adapted to the high spectral dimension of hyperspectral images is discussed in this article. The kernel is based on the Mahalanobis distance defined for each class. A parsimonious probabilistic model is used for the inversion of the covariance matrix. The kernel reaches a trade-off between a conventional Gaussian kernel and a Gaussian kernel on the first principal components of the considered class. The proposed approach is compared to other methods for the classification of Ash tree in hyperspectral images. In terms of accuracies, the proposed kernel performs better than conventional Mahalanobis kernel and performs as well as the conventional Gaussian kernel.Un noyau adapté à la grande dimension spectrale des imag...
L'imagerie hyperspectrale, grâce à un nombre élevé de bandes spectrales très fines et contigües, est...
For the classification of hyperspectral images (HSIs), this paper presents a novel framework to effe...
International audienceA family of parsimonious Gaussian process models for classification is propose...
National audienceA kernel adapted to the high spectral dimension of hyperspectral images is discusse...
National audienceA kernel adapted to the high spectral dimension of hyperspectral images is discusse...
The definition of the Mahalanobis kernel for the classification of hyperspectral remote sensing imag...
International audienceThe definition of the Mahalanobis kernel for the classification of hyperspectr...
International audienceThe definition of the Mahalanobis kernel for the classification of hyperspectr...
International audienceThe definition of the Mahalanobis kernel for the classification of hyperspectr...
The classification of high dimensional data with kernel methods is considered in this article. Explo...
International audienceThe classification of high dimensional data with kernel methods is considered i...
Cette thèse est consacrée à l'analyse statistique de données en grande dimension. Nous nous intéress...
Dans cette thèse, nous proposons et développons des nouvelles méthodes et algorithmes spectro-spatia...
La finesse de la résolution spectrale et spatiale des images hyperspectrales en font des données de ...
L'imagerie hyperspectrale, grâce à un nombre élevé de bandes spectrales très fines et contigües, est...
L'imagerie hyperspectrale, grâce à un nombre élevé de bandes spectrales très fines et contigües, est...
For the classification of hyperspectral images (HSIs), this paper presents a novel framework to effe...
International audienceA family of parsimonious Gaussian process models for classification is propose...
National audienceA kernel adapted to the high spectral dimension of hyperspectral images is discusse...
National audienceA kernel adapted to the high spectral dimension of hyperspectral images is discusse...
The definition of the Mahalanobis kernel for the classification of hyperspectral remote sensing imag...
International audienceThe definition of the Mahalanobis kernel for the classification of hyperspectr...
International audienceThe definition of the Mahalanobis kernel for the classification of hyperspectr...
International audienceThe definition of the Mahalanobis kernel for the classification of hyperspectr...
The classification of high dimensional data with kernel methods is considered in this article. Explo...
International audienceThe classification of high dimensional data with kernel methods is considered i...
Cette thèse est consacrée à l'analyse statistique de données en grande dimension. Nous nous intéress...
Dans cette thèse, nous proposons et développons des nouvelles méthodes et algorithmes spectro-spatia...
La finesse de la résolution spectrale et spatiale des images hyperspectrales en font des données de ...
L'imagerie hyperspectrale, grâce à un nombre élevé de bandes spectrales très fines et contigües, est...
L'imagerie hyperspectrale, grâce à un nombre élevé de bandes spectrales très fines et contigües, est...
For the classification of hyperspectral images (HSIs), this paper presents a novel framework to effe...
International audienceA family of parsimonious Gaussian process models for classification is propose...