Real-World ISAR Object Recognition and Relation Discovery Using Deep Relation Graph Learning

REMI: Few-Shot ISAR Target Classification Via Robust Embedding and Manifold Inference

Unknown image deformation and few-shot issues have posed significant challenges to inverse synthetic aperture radar (ISAR) target classification. To achieve robust feature representation and precise correlation modeling, this article proposes a novel two-stage few-shot ISAR classification network, dubbed as robust embedding and manifold inference (REMI). In the robust embedding stage, a multihead spatial transformation network (MH-STN) is designed to adjust unknown image deformations from multiple perspectives. Then, the grouped embedding network (GEN) integrates and compresses diverse information by grouped feature extraction, intermediate feature fusion, and global feature embedding. In the manifold inference stage, a masked Gaussian graph attention network (MG-GAT) is devised to capture the irregular manifold of samples in the embedding space. In particular, the node features are described by Gaussian distributions, with interactions guided by the masked attention mechanism. Experimental results on two ISAR datasets demonstrate that REMI significantly improves the performance of few-shot classification and exhibits robustness in various scenarios.

Real-World ISAR Object Recognition Using Deep Multimodal Relation Learning

Real-world inverse synthetic aperture radar (ISAR) object recognition is a critical and challenging problem in computer vision tasks. In this article, an efficient real-world ISAR object recognition method is proposed, namely, real-world ISAR object recognition (RIOR), based on deep multimodal relation learning (DMRL). It cannot only handle the complex multimodal recognition problem efficiently but also exploit the relations among the features, attributes, labels, and classes with semantic knowledge: 1) an adaptive multimodal mechanism (AMM) is proposed in convolutional neural network (CNN) to substantially promote the CNN sampling and transformation capability and significantly raise the output feature map resolutions by keeping almost all of the information; 2) deep attribute relation graph learning (DARGL) is proposed to jointly estimate the large numbers of heterogeneous attributes and collaboratively explore the relations among the features, attributes, labels, and classes with common knowledge graphs; and 3) relational-regularized convolutional sparse learning (RCSL) is proposed to further achieve good translation invariance and improve the accuracy and speed of the entire system. Extensive qualitative and quantitative experiments are performed on two real-world ISAR datasets, demonstrating that RIOR outperforms the state-of-the-art methods while running quickly.