RFI Detection in X-Band SAR Imagery using Machine Learning

Abstract

The present study investigates and develops an automated technique for detecting Radio Frequency Interference (RFI) in X-band Synthetic Aperture Radar (SAR) Single-Look Complex (SLC) images. RFI patterns significantly reduce the quality and utility of SAR images. Thus, efficient RFI detection is essential for selecting SAR images without significant RFI for SAR applications. To this end, we utilized multi-year imagery acquired by TerraSAR-X and TanDEM-X, previously labeled in a prior study, to curate datasets containing both RFI and non-RFI images. Given our dataset's limited amount of RFI-labeled data, we focused on exploring various techniques to address this challenge and achieve reliable detection. We propose conducting comparative analyses of the performance of image augmentation techniques and SAR-image-tailored data augmentation methods for the RFI dataset. The main objective is to develop a machine learning model tailored to this problem and leverage these novel augmentation strategies to overcome this limitation.