Classification of Polarimetric-SAR data with neural network using combined features extracted from scattering models and texture analysis

This paper shows a study on an alternative method for classification of polarimetric-SAR data. The method is designed by integrating the combined features extracted from two scattering models (ie., Freeman decomposition model and Cloude decomposition model) and textural analysis with distribution-free neural network classifier. The neural network classifier (wich is based on a feed- forward back-propagation neural network architecture) properly exploits the information in the combined features for providing high accuracy classification results. The effectiveness of the proposed method is demonstrated using E-SAR polarimetric data acquired on the area of Penajam, East Kalimantan, Indonesia.