Fuqiang Zeng

Ship detection in synthetic aperture radar (SAR) imagery plays an important role in maritime surveillance and security applications. However, in complex near-shore and port scenes, strong background clutter, significant variations in target scale, and the need for efficient feature representation make it challenging to achieve a good balance between detection accuracy and computational efficiency. To address these challenges, this paper proposes a lightweight oriented bounding box (OBB)-based SAR ship detection network based on feature enhancement and cross-scale fusion. By systematically redesigning key components, including feature enhancement, downsampling modeling, multi-scale fusion, and oriented target representation, the proposed method achieves coordinated optimization of accuracy and efficiency in complex scenarios. Specifically, a Multi-Scale Coordinate-Channel Attention mechanism (MSCCA) is constructed to enhance discriminative ship features while suppressing background interference. An Adaptive Dual-pooling Fusion Downsampling strategy (ADFD) is then introduced to reduce spatial resolution while alleviating fine-grained feature loss. In addition, a Lightweight Adaptive Multi-Scale Feature Enhancement module (LAMFE) is designed to improve multi-scale modeling capability under strict computational constraints. Finally, a Lightweight Oriented Bounding Box detection head (LwOBB) is employed to enable accurate localization of ship targets with arbitrary orientations. The experimental results on the publicly available R-SSDD and RSDD-SAR datasets show that the proposed detection network achieves higher detection accuracy while maintaining a lightweight structure. The model contains only 2.06 M parameters and has a computational complexity of 5.7 GFLOPs. Compared with several mainstream detection methods, the proposed method has achieved significant improvements in detection accuracy and effectively reduces the occurrence of missed detections and false alarms.

Vessel trajectory prediction tasks currently face several challenges, including the diversity of behavioral patterns, interference from anomalous data, and limited generalization capabilities of existing models. These challenges are particularly pronounced when dealing with multi-pattern trajectory characteristics, where achieving both high prediction accuracy and robust cross-scenario adaptability remains difficult. To address these issues, this paper proposes a high-precision vessel trajectory prediction framework. Firstly, the local outlier factor algorithm is applied to eliminate anomalies from raw automatic identification system data, thereby improving data quality. The Douglas–Peucker simplification and the Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) trajectory clustering algorithm are then used to construct datasets characterized by different turning behaviors. Subsequently, a prediction model called local-causal temporal attention network (LCTA-Net) is developed, based on a pruned Transformer encoder. LCTA-Net incorporates a temporal residual encoding module and a causal feature-augmented attention mechanism. These components jointly improve the model’s capacity to model temporal dependencies within trajectory sequences. In addition, a fixed-window local neighborhood search strategy is introduced to improve the spatial continuity and physical feasibility of the predicted trajectories. Finally, experimental results on two constructed trajectory datasets demonstrate that the proposed method significantly outperforms several state-of-the-art models across multiple error metrics, confirming its superior prediction accuracy and strong cross-scenario generalization capability.