New kernel methods increase remote sensing accuracy

AUGUST 27, 2008--The support vector machine (SVM) kernel method has been successfully used in hyperspectral image classification. Nevertheless, SVMs must be adapted to the specific needs of the field.

Aug 27th, 2008

AUGUST 27, 2008--The support vector machine (SVM) kernel method has been successfully used in hyperspectral image classification. Nevertheless, SVMs must be adapted to the specific needs of the field. Inclusion of contextual information in the classifier is necessary to produce more spatially homogeneous classification maps. Multisensor and multitemporal information has been also synergetically combined with kernels. Another kernel method, the one-class SVM, is aimed at identifying samples of one particular class and rejecting the others. The method was originally introduced for anomaly detection then used for dealing with incomplete and unreliable training data and recently reformulated for change detection. Lately, semi-supervised kernel-based classifiers--for example, the transductive and the Laplacian--have been introduced to exploit the wealth of unlabeled data in the image.

For more information, go to http://spie.org/x25547.xml?highlight=x2420.

More in Defense & Aerospace