SAR Target Data Generation Method with Integrated Scattering Features

Synthetic Aperture Radar (SAR) is widely used in military and civilian fields. However, the difficulty in obtaining high-quality measured data limits the availability of existing SAR vehicle target datasets. As a result, deep learning-based data generation is frequently utilized for SAR target augmentation because of its affordability and simplicity of use. For SAR targets, scattering features are the most important information that discriminates the targets. However, the state-of-the-art methods do not adequately account for the target scattering information in SAR data generation. Furthermore, the scattering feature distribution of the generated data differs much from that of the real SAR data, making it difficult to effectively represent the scattering features of the target. Aiming at the above problems, we propose a SAR vehicle target generation method that fuses scattering features. The attributed scattering center (ASC) of the target is extracted by the orthogonal matching pursuit (OMP) method, which is computationally efficient. Then, a multidimensional feature representation is constructed by combining the original SAR data, fully utilizing the scattering features of SAR images. We improve the CycleGAN by adding a converter module, so that the network model can learn more information in the multidimensional features. It aims to enhance the ability of model to capture and learn the scattering mechanisms of SAR targets, enabling the generation of data that closely aligns with these scattering features. Finally, to validate the effectiveness of our method, extensive experiments are conducted using the MSTAR and SAMPLE datasets, demonstrating its superiority over existing techniques. The proposed method can supplement SAMPLE dataset by filling in missing target data, thereby enriching the dataset with additional target categories and increasing its sample amount. At the same time, the proposed method provides a new direction for the high-quality generation of SAR ship and aircraft targets.