Multi-resolution phase model reconstruction for phase unwrapping

An algorithm for phase unwrapping is presented that deals with the problem of error propagation due to interferometric noise. The proposed technique belongs to the category of global analysis methods, and it is based on the generation of an approximate phase model, which is iteratively refined until a stopping condition is reached. The underlying principle is that the interferogram is partitioned into square-shaped, non-overlapping elements such that the unwrapped phase on each element can be modelled by the mean value of the local slopes. Starting from this information, which in two dimensions is represented by flat surfaces that best fit the real ground behaviour, the global model is reconstructed by linking the vertices of the found planes through the solution of a least mean square problem. The effectiveness of this method is limited by the presence, in real interferometrical data, of large areas in which the information about local slopes is absent due to layover, noise, vegetation, etc. In order to solve this problem, a weighted version of the algorithm has been developed.