A New Airborne Radar Target Detection Approach Based on Conditional Generative Adversarial Nets

To solve the degradation of clutter suppression effect of traditional radar signal processing method in the complex battlefield environment, especially for the airborne radar with more complex clutter background, in this paper, we use Conditional Generative Adversarial Nets (CGAN) to train a generator which can find interested targets in a complex clutter distribution. It exploits the distribution of interested targets as supervised information so as to assist the extraction of high-resolution features. Moreover, for the time-varying battlefield environment, only by updating model parameters can our method adapt the change of distribution of radar echo data. The effect of our method on clutter suppression is validated based on the experiments on the radar datasets.