A Deep Learning Method for SAR Imaging in Sparse Field

Motivated by the successes of data driven approaches in various fields, a great number of deep learning methods have been applied to SAR images. In this work, we propose a deep neural network named RG-Net, which can generate SAR images from raw echo data in the sparse target domain. Experiments are carried out on a simulation dataset, and the results show the superiority of the proposed method on imaging quality and computational efficiency. The source code is at https://github.com/zhangyingerjelly/RG-Net