Two-dimensional wavelet algorithm for interferometric synthetic aperture radar phase filtering enhancement

Abstract

Interferometric synthetic aperture radar (InSAR) coherence is valuable information used for quantifying the degree of accuracy of the interferometric phase (interferogram). We propose two new algorithms for InSAR filtering based on the coherence map. These are complementary algorithms for accurate InSAR phase filtering in the wavelet domain. The first algorithm is an enhanced version of the Lopez and Fabregas filter, called wavelet interferometric phase filter (WInPF), where the main improvement is in the noise mask growing step. The new algorithm overcomes the limited accuracy of noise mask computation used in the “Filtrage par Approche Multiechelle Modifie” (FAMM) algorithm by giving more accurate noise mask values with respect to the InSAR coherence information. The second approach proposed in this paper looks at reducing the impulse noise still tainting the SAR interferogram after operating the WInPF or the FAMM. Thus, we propose an improved version of the adaptive switching median filter dedicated to the case of the interferograms, which takes into account the corresponding coherence map values. The proposed approaches are validated on simulated interferograms and tested on real InSAR data acquired with the Radarsat-2 and Envisat satellites over Mahdia in Tunisia (North Africa) and Mount Etna in Italy (Europe).