Application of Generative Adversarial Networks for GBSAR Data Synthesis

Ground-Based Synthetic Aperture Radar (GBSAR) is crucial for applications like geological surveys and infrastructure monitoring, but data acquisition is often time-consuming and expensive. This paper explores using Generative Adversarial Networks (GANs) to synthesize GBSAR data, addressing data scarcity issues. Using data from a 24 GHz GBSAR system, we demonstrate that GANs can generate realistic synthetic datasets, thereby augmenting the training process and improving the accuracy of deep learning models. Our findings indicate that GANs effectively expand dataset size and lead to enhancements in classification model performance, offering a cost-effective supplement to GBSAR data acquisition.