Abstract
Satellite Image Time Series (SITS) are a very useful source of information for geoscientists especially for land cover monitoring. In this paper a new temporal classification approach for High Resolution (HR) SITS is proposed. It suggests a Bayesian combination between a pixel and a region, SVM (Support Vector Machine) based techniques where SVM is considered as a probabilistic classifier. RBF (Radial Basis Function) based SVM kernel is used to classify pixel evolution while a graph based SVM kernel is considered to analyze region temporal behavior.