A maximum likelihood a priori filter for interferometric phase

Image regions of low coherence represent a major problem in the evaluation of InSAR phases. One possibility to deal with this problem consists in filtering the wrapped phase. The interferometric wrapped phase of InSAR images can be interpreted as a random variable. Its well known pdf has the wrapped topographic phase as parameter. Therefore it is possible to estimate the topographic wrapped phase by using a maximum likelihood approach based on a locally affine model. The paper gives a description of the algorithm and presents tests on a synthetic as well as on a real scene