Extracting Residential Area Information from Dual-SAR Image Based on Object-Oriented Technique

Abstract

Quick and accurate extracting residential area information has important practical significance to plantation protection, city planning, and reasonable land use. Hence the paper introduces an innovative method about extracting residential area from dual- polarization SAR image. First of All, create area object using region grow algorithm, secondly select residential area samples artificially and compute average covariance matrix of each type, then extract residential area objects based on Wishart distance. According to the above theory, design the corresponding procedure, and do experiment on dual- polarization SAR image of a region, the result indicates that the method for residential area information extraction has higher precision, less calculations, faster speed than the existing ones.