Accuracy improvement of maximum likelihood inversion of forest height with PolInSAR

Polarimetric synthetic aperture radar interferometry (PolInSAR) technique has been intensively used for extracting forest heights. Among the forest height inversion algorithms, maximum-likelihood estimate computes the maximum-likelihood estimate of the vegetation parameters based directly on the sample coherency matrix. The estimation of the sample coherency matrix is of great importance in this inversion process. A coherency matrix estimation method with adaptive averaging window derived from coherence region shape parameters is presented. Then the estimated coherency matrices are applied in the maximum likelihood inversion algorithm. In order to confirm the validity of the proposed method, simulated L-band PolInSAR data provided by ESA is used. Experiment results suggest that this method can help improve the accuracy of maximum likelihood inversion of forest heights.