Journal of Computational Information Systems 6:13 (2010) 4417-4425 Available at http://www.Jofcis.com 1553-9105/ Copyright © 2010 Binary Information Press December, 2010 PSONN Used for Remote-Sensing Image Classification Yudong ZHANG † , Shuihua WANG, Lenan WU, Yuankai HUO School of Information Science and Engineering, Southeast University Abstract This section proposes a hybrid classifier for polarimetric synthetic aperture radar (SAR) images. The feature sets consisted of span image, the H/A/α decomposition, and the gray-level co-occurrence matrix (GLCM) based texture features. Then, the features were reduced by principle component analysis (PCA). Early stop (ES) was adopted to prevent overfitting. Finally, a 3-layer neural network (NN) was constructed, and particle swarm optimization (PSO) was employed to fasten the learning. The results on Flevoland sites compared to adaptive back-propagation (ABP) neural network demonstrated the validness and superiority of our method in terms of confusion matrix and overall accuracy. Keyword: Image Classification; Forward Neural Network; Particle Swarm Optimization; Adaptive Back-propagation. 1. Introduction The classification of different objects, as well as different terrain characteristics, with single channel monopolarisation SAR images can carry a significant mount of error even when operating after multilooking [1]. One of the most challenging applications of polarimetry in remote sensing is landcover classification using fully polarimetric SAR (PolSAR) images. The Wishart maximum likelihood (WML) method has often been used for PolSAR classification [2]. However, it does not take explicitly into consideration the phase information within polarimetric data. Furthermore, the covariance or coherency matrices are determined after spatial averaging and therefore can describe only stochastic scattering processes while certain objects, such as man-made objects, are better characterized at pixel-level [3]. To overcome above shortcomings, polarimetric decompositions were introduced with an aim at establishing a correspondence between the physical characteristics of the considered areas and the observed scattering mechanisms. The most effective method is the Cloude decomposition, also known as H/A/α method [4]. Recently, texture information has been extracted, and used as a parameter to enhance the classification results. The gray-level co-occurrence matrices (GLCM) were already successfully applied to classification problems. We choose the combination of H/A/α and GLCM as the parameter set of our method. The next problem is how to choose the best classifier. In the past years, standard multi-layered feed-forward NNs with a back propagation (BP) algorithm have been applied for SAR image classification [5]. BPs are † Corresponding author. Email addresses: zhangyudongnuaa@gmail.com (Yudong ZHANG), shuihuaw2007@gmail.com (Shuihua WANG), wuln@seu.edu.cn (Lenan WU)