RFI detection and mitigation in SAR data

Radio Frequency Interference (RFI) are incoherent electromagnetic interference signals that appear as various kind of bright linear features in azimuth time range frequency diagram. RFI causes significant reduction in image contrast and introduces artifacts in image that may affect the polarimetric signatures and degrades the overall radiometric quality. Although the matched filtering which is an integral part of SAR data processing system has an inherent ability of interference suppression, yet high power RFI is still visible in Azimuth time Range frequency domain as bright stripes. It also introduces hazyiness in the image. The technique presented in this paper is a modified variant of an existing approach retaining the basic philosophy for RFI detection. It uses hypothesis testing based on z-test combined with Tukey biweights to estimate robust statistics for the purpose of RFI detection. Denoising of RFI mask is performed using statistical information of data. RFI correction is further performed using interpolation in magnitude based on local statistics.