The Presentation of an Algorithm for Interference Detection in the Synthetic Aperture Radar

The synthetic aperture radar is an imaging radar that has a high resolution. The syntheticaperture radar image may be degraded by the interference of radio frequencies and anincomprehensible image may be created. Interferences in the synthetic aperture radars aredivided into the three categories of , , and , which represent radio frequency noise interference,narrow band interference and wideband interference, respectively. To effectively reduce theinterference in synthetic aperture radar images, first the presence of interference and its typeshould be asserted and then the interference reduction algorithms should be calculated accordingto interference type. In this paper an algorithm for the detection of interference and its type inthe synthetic aperture radar images is presented. Whilst in the previous articles the SSD methodis used for interference detection, in this paper we have used the Faster RCNN method based onneural network convolutional which has a higher speed and accuracy than the SSD method. Inthis method, first a neural network is trained with the ability of multiple classification. Then theFaster RCNN is constructed with the neural network and and is trained by 25 time - frequencyimages from the artificial aperture radar signal. The trained network is able to detect anyinterference in the radar signal of a synthetic window with 99% accuracy. After detecting theinterference by the proposed algorithm, the normalized least mean square filter is able to reducethe interference and improve the radar image. This filter operates similarly in decreasing allthree types of interference.