Minimization of the ambiguity of merging of urban builtup and fallow land features by generating ‘C2’ covariance matrix using spaceborne bistatic dual Pol SAR data

SAR data due to its all-weather availability and Polarimetrie properties helps to delineate urban built up areas from fallow lands which otherwise show similar signatures and get merged in an optical dataset. This makes their classification difficult. The problem becomes more acute in months of April when there is no standing crop in fields and most of agricultural fields are lying fallow and show similar tone as surrounding urban areas. To solve this ambiguity, in this study Polarimetric `C2' Covariance matrix was generated and its various components helped to delineate urban areas from fallow lands. This study makes use of `C2' Covariance matrix which is a 2×2 matrix and dual pol dataset from TerraSAR-X having HH and HV Polarisation channels. Further a Support Vector Machine classification was performed on the RGB generated from the C2 matrix, using Radial basis function kernel and it showed better classification as compared to SVM Classified optical imagery thereby minimizing the ambiguity of merging of urban and fallow land features.