BAYESIAN PROBABILISTIC INFORMATION PROCESSING FOR PHASE UNWRAPPING USING SAR INTERFEROMETRY DUE TO CONJUGATE GRADIENT METHOD

We investigate wave-front reconstruction in remote sensing using the synthetic aperture radar (SAR) interferometry on the basis of Bayesian inference both using the method of maximum entropy (ME) and the Maximum A Posteriori (MAP) estimation due to the conjugate gradient method (CGM). From a theoretical viewpoint, we first confirm that the method of ME is successful in phase unwrapping with high degree of accuracy around the Bayes-optimal condition, using Monte Carlo simulations for a set of wave-fronts generated by the assumed true prior enhancing smooth structures in the pattern of wave-fronts. Next, from the practical viewpoint, we utilized the MAP estimation using the CGM to construct a practical and useful method for phase unwrapping in remote sensing using SAR interferometry. Using numerical simulations, we clarify that wave-fronts are perfectly reconstructed without using the model prior, if original wave-fronts are not corrupted by any noises, even if aliasing occurs in the optical measurement, and also that wave-fronts were accurately reconstructed using appropriate model of the true prior, if the original wave-fronts were corrupted by some noises