Performance analysis of different decomposition methods for classification of crops and other land cover targets using Risat-1 hybrid SAR imageries

The purpose of the current study is to analyze the decomposition methods for detecting crops and other land-cover targets based on Risat-1 hybrid polarimetric Synthetic Aperture Radar (SAR) imageries. In this study, the scattering mechanisms of every target feature have been investigated in terms of odd bounce, even bounce and volume component using m-χ, m-δ, m-α and modified m-χ decomposition methods to understand their performances in a better way. The comparative analysis of four decomposition methods has been performed in two test site areas namely Burdwan in the state of West Bengal and Bharatpur in the state of Rajasthan. Modified m-χ decomposition method provides the better results followed by conventional m-χ decomposition for segmentation of target features. Other conventional methods such as m-δ, m- α decomposition methods are in decreasing order of performance. The volume component is still found over-estimated in settlement using the modified m-χ decomposition method.