.png){kind=icon}
.png){kind=icon}
Return
SAR Missing Echo Imaging via Hierarchical Learning Deployed on Hybrid Network
.png){kind=icon}
PRE
AI
DOI：10.1109/TAES.2025.3615194.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The fine quality of synthetic aperture radar (SAR) image has made various kinds of downstream tasks feasible. Yet, the received echoes suffer from random missing circumstances in practical. In these conditions, the imaging quality is degraded, thereby reducing the performance of downstream tasks. To address these problems, a learnable SAR imaging method is proposed in this article. Different from previous works, more attention is paid to generating high-quality images from random missing echoes. First, a filling and enhancing imaging network (FEINet) is developed, which is a hybrid network. The advantages of convolution and multihead attention (MHA) mechanism are combined for SAR random missing echo imaging. Second, a hierarchical learning strategy called the deep supervision learning strategy is presented. In this strategy, the filling imaging network and the enhancing imaging network are trained jointly. A joint loss measurement is then presented to guide the model to converge. Third, the converged model is used to generate high-quality single-look complex images from random missing echoes. Multiple rounds of experiments are performed. The results prove that the proposed learning strategy plays an important role in the performance of FEINet. Furthermore, the FEINet achieves state-of-the-art imaging performance among all the compared methods.
Keywords:
Deep supervision learning strategy
random missing echoes
synthetic aperture radar imaging
Journal
IF:
5.7
Papers:
676
Citations:
2.4W