Fast Filtering of Interferometric Phase Noise in Insar Data

Interferometry is used to obtain information about the Earth's surface elevation. Surface height can be derived from inter-ferometric phase images, although the inherent phase noise reduces the accuracy and reliability of this information. Phase noise poses a challenge in obtaining precise information, es-pecially in steep areas. Therefore, it is essential to reduce the noise before extracting the surface information embed-ded in the interferometric phase. Widely used phase filtering methods can be grouped into three categories: spatial domain methods, transform domain methods, and nonlocal methods (NL). The Refined Lee Filter is a spatial domain method that reduces noise through local statistics. Instead of adopting the conventional approach of using a moving square window, the Refined Lee Filter employs sixteen fringe-aligned windows and develops an algorithm to select the window containing pixels with the most homogeneous phases. This study presents a proposal for filtering interferometric phase based on a numerical expression for multilook phase difference, which is more manageable compared to the expression used in the Refined Lee Filter. The proposed filter aims to improve the execution time in the phase noise filtering process. To achieve this goal, we introduce a constraint on the number of looks into the phase difference expression defined by Gierull [4], with the hope of making the filtering process faster. Nu-merical experiments on simulated and real data were used to validate the proposed approach.