SAR-GTR: Attributed Scattering Information Guided SAR Graph Transformer Recognition Algorithm for Consumer-Grade Low-Altitude Economy

作者：Xuying Xiong, Xinyu Zhang, Weidong Jiang, Li Liu, Yongxiang Liu, Tianpeng Liu · 发表于：IEEE Transactions on Consumer Electronics · 年份：2025 · DOI：10.1109/tce.2025.3586944 · 被引用次数：9 · 研究领域：E-commerce and Technology Innovations

With the rapid development of consumer-grade Uncrewed Aerial Systems (UAS) and the Internet of Things (IoT), Synthetic Aperture Radar (SAR) provides important support for multiple UASs to realize target detection and recognition through the IoT. Utilizing electromagnetic scattering information for SAR data interpretation is the current research focus in the field. Graph Neural Networks (GNNs) can effectively integrate physical and human prior knowledge and are lightweight, making them ideal for deployment in edge-level consumer products. In this study, we delve into the electromagnetic backscatter information of single-channel SAR and revisit the limitations of GNNs in SAR interpretation and propose the Synthetic Aperture Radar Graph Transformer Recognition Algorithm (SAR-GTR). SAR-GTR avoids confusion and loss of information by distinguishing the discrete and continuous parameters in the mapping method. In addition, SAR-GTR introduces an edge information enhancement channel to facilitate interactive learning of node and edge features to capture robust and global structural features of the target. Meanwhile, SAR-GTR fully utilizes the hierarchical structural information of the target through global node encoding and edge position encoding. In brand-new scenarios such as consumer-grade UAS cooperative operation and IoT target recognition, SAR-GTR can adapt to different recognition angles and maintain the lightweight characteristics to satisfy the needs of low-altitude economy ...