LEICI 25638
INSTITUTO DE INVESTIGACIONES EN ELECTRONICA, CONTROL Y PROCESAMIENTO DE SEÑALES
Unidad Ejecutora - UE
artículos
Título:
Unsupervised Polarimetric SAR Image Classification using G0,p Mixture Model
Autor/es:
J.I. FERNÁNDEZ MICHELLI; C. H. MURAVCHIK; M. HURTADO; J.A. ARETA
Revista:
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS
Editorial:
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
Referencias:
Año: 2017 vol. 14 p. 754 - 758
ISSN:
1545-598X
Resumen:
Abstract?This letter proposes a polarimetric Synthetic ApertureRadar (SAR) image classification method based on theExpectation-Maximization (EM) algorithm. It is an unsupervisedalgorithm that determines the number of classes in the scenefollowing a top-down strategy using a covariance based hipothesistest. A G0,p mixture model is used to describe Multilook Complex(MLC) polarimetric data, and the proposed algorithm is tested insimulated and real datasets obtaining good results. The classificationperformance is evaluated by means of the overall accuracyand the kappa indices obtained from Montecarlo analysis. Finally,the results are compared to those obtained by other classic andrecently developed classification algorithms.
