Despeckling of Multifrequency SAR Data Using Electromagnetic Scattering Models
Gerardo Di Martino, Alessio Di Simone, Antonio Iodice, Daniele Riccio, Giuseppe Ruello · 2023
Interpretation and processing of synthetic aperture radar (SAR) imagery is negatively affected by speckle noise, that can be faced by proper despeckling pre-processing. In this work, we propose a simple approach for the despeckling of multi-frequency SAR data under the hypotheses that they are relevant to bare soil surfaces and that are acquired by the same sensor, e.g., AIRSAR. The proposed approach is based on a frequency-compensation step, where the dependence of the signal strength upon operating frequency is compensated by means of surface scattering model suited to natural surfaces. Once such a pre-processing step is carried out, any despeckling filter suited to SAR time series filtering can be applied. Quantitative indicators evaluated on both simulated and actual SAR data reveal the benefits of the proposed compensation step w.r.t. pure multitemporal and single-channel filtering.