A critical analysis of EM based fusion of different polarization data for effect on land cover classification

Abstract

Land cover classification is an important activity in social, economical, geographical, ecological and risk planning of a country. Production of accurate land cover maps autonomously is still a challenging problem. It motivates the study and evolution of methods to tackle this problem. The purpose of the present study is to critically analyze the effects of image fusion on various land cover classification. The importance of using different polarizations of the single SAR data has been emphasized. A simplified EM algorithm for fusion of different permutations and combinations of multi-polarized PALSAR data is presented through modeling that is valid for Gaussian as well as non-Gaussian distortions. K-means unsupervised algorithm has been applied for the classification of various land cover i.e. water, urban, wetland, baresoil, short vegetation and tall vegetation. The proposed method is intuitive and simple in that it uses PALSAR data after pre-processing straightaway without any transformations and also estimates the missing information in each of the channel without any a priori knowledge. A critical analysis of fusion effects on different land covers is presented on the basis of various accuracies. This type of study will be helpful in further enhancing accuracy of land cover maps minimizing human intervention.