A new adaptive multiresolution noise-filtering approach for SAR interferometric phase images

This letter presents a novel adaptive multiresolution technique to estimate the local fringe frequency in synthetic aperture radar (SAR) interferometric phase images. According to the coherent summation principle, the size and the shape of the interferogram spatial resolution for local frequency estimation are adapted pixel by pixel to track the topographic changes and optimize the space-frequency resolution. We show that, comparing conventional filters with the processing window of a fixed size (such as 7 x 7) and shape (such as a square), the proposed approach avoids the loss of phase fringes and provides a more accurate representation of the scene topography. In this letter, we present a method to eliminate invalid frequency estimation. A fast frequency estimation approach modified from the method proposed by Trouve et al. is also shown. Finally, both simulated data and actual SAR data are utilized to illustrate the effectiveness of the proposed method.