Parameter Estimation of Multilook Polarimetric SAR Data Based on Fractional Determinant Moments

Abstract

This letter presents a novel method to estimate the equivalent number of looks (ENLs) and the texture parameters when multilook polarimetric SAR data are stochastically modeled by the product model. The method is based on the fractional moments of the determinant of the multilook polarimetric covariance (FMDC) matrix. The estimator of ENL is independent of the distribution of the texture model, and the estimator of the texture parameters can be derived for all commonly used multilook PolSAR distributions. $\\mathcal {K}_{d}$ is given in this letter as multilook PolSAR distribution. The performance of the FMDC estimator is compared to the performance of other known estimators using simulated and real data.