Optimum Ensemble Classification For Fully Polarimetric Synthetic Aperture Radar Data Using Global-Local Classification Approach

In this paper, a ensemble classification for fully polarimetric synthetic aperture radar (polsar) data using a global-local classification approach is discussed. in the first step, to perform the global classification, the training feature space is divided into a specified number of clusters. in the next step local classification over each of these clusters is conducted; which contains elements of several classes, a base classifier. thus, an ensemble of classifiers was formed; each of them professionally acts as a part of the feature space. to achieve more diversity, the data set is independently partitioned into a variable number of clusters by h/α classifier and k-means algorithm. to combine the outputs of different arrangements, majority voting, naïve bayes and a heuristic combination rule by taking into account the classification accuracy and reliability (which in polsar classification less attention has been paid to it) as objective functions were used. the experimental results over two polsar images preved effectiveness of the proposed algorithms in compare to baseline methods.