A method for radar time-frequency image data suppression based on gradient-weighted class activation mapping

In performing the radar target detection task under the background of sea clutter, bad weather conditions produce strong sea clutter that is not favorable to radar target detection, and it is easy to produce model mismatches and false alarms and other unfavorable detection conditions when performing the detection. This paper proposes a radar time-frequency image data suppression method based on gradient-weighted class-activation mapping, which can better complete the suppression of sea clutter and provide a better precondition for the subsequent radar target detection. A good suppression effect can be achieved on both types of real radar datasets, which proves the effectiveness of the present suppression method for suppression in various scenarios.