Track interruption will cause serious interference to target tracking, situation awareness, and information fusion. And it is often triggered by a variety of reasons, such as target maneuver, the blind spot of Doppler radar, and excessive long sampling interval, among other causes. All the existing algorithms are based on the hypothetical target motion models and adopt the theory of statistical estimation to measure the similarity between predicted tracks to associate, which have some disadvantages such as unreasonable hypotheses, inapplicable models, and uncertain thresholds. In order to address these problems, in this article, we propose a track segment association dual contrast neural network (TSADCNN) to complete track segment association tasks. First, based on deep metric learning, we propose a temporal and spatial information extraction module to extract track information, especially symmetric correlation information and raise dimension. Second, by constructing an appropriate optimization function, TSADCNN makes track segments belonging to the same target closer, track segments belonging to different targets farther in high-dimensional space. Finally, TSADCNN will pick the nearest neighbor vectors in high-dimensional space as associated tracks. Extensive experiments on simulation data are conducted to evaluate the effectiveness of TSADCNN, and the results demonstrate that this method has good adaptability and anti-noise ability. Furthermore, the contrast experiment shows that it can outperform other TSA methods in both quality and efficiency. Compared with the best performance, the average true association rate of TSADCNN can be increased by 1.3% in long interrupt intervals.

Traditional Track Segment Association (TSA) methods used track position vectors or other track information to get association results. However, simply extracting the track position information to form track vectors will lead to the loss of irregular structure information. To solve this problem, a Track Graph Representation Association (TGRA) method is proposed. Through node‐level local track point embedding and graph‐level track graph embedding, representations of track segments can be obtained and irregular structure information can be retained simultaneously. Then, by the constraint of loss function, track segments belonging to the same target become closer, track segments belonging to different targets become farther in representation space. Finally, the nearest neighbour embedding in representation space is picked as associated tracks. Simulation results demonstrate that TGRA has good adaptability and anti‐noise ability, and it can outperform other TSA methods in both quality and efficiency. Compared with the best performance, the average true association rate of TGRA can be increased by 2.8% in short interrupt intervals and 1.3% in long interrupt intervals, respectively.

Ship target identification is of great significance in both military and civilian fields. Many methods have been proposed to identify the targets using tracks information. However, most of existing studies can only identify two or three types of targets, and the accuracy of identification needs to be further improved. Meanwhile, they do not provide a reliable probability of the identification result under a high-noise environment. To address these issues, a Bayesian-Transformer Neural Network (BTNN) is proposed to complete the ship target identification task using tracks information. The aim of the research is improving the ability of ship target identification to enhance the maritime situation awareness and strengthen the protection of maritime traffic safety. Firstly, a Bayesian-Transformer Encoder (BTE) module that contains four different Bayesian-Transformer Encoders is used to extract discriminate features of tracks. Then, a Bayesian fully connected layer and a SoftMax layer complete the classification. Benefiting from the superiority of the Bayesian neural network, BTNN can provide a reliable probability of the result, which captures both aleatoric uncertainty and epistemic uncertainty. The experiments show that the proposed method can successfully identify nine types of ship targets. Compared with traditional methods, the identification accuracy of BTNN increases by 3.8% from 90.16%. In addition, compared with non-Bayesian Transformer Neural Network, the BTNN can provide a more reliable probability of the identification result under a high-noise environment.

Ship target recognition based on trajectories has great potential in the field of target recognition. In the existing research, the context information is ignored, which limits the improvement of ship target recognition ability. In addition, the process of trajectory feature extraction is complex, and recognition accuracy needs to be further improved. In this paper, a ship target recognition method based on a context-enhanced trajectory is proposed. The maritime context knowledge base is constructed to enhance the trajectory information and to improve the separability of different types of target trajectories. A deep learning model is used to extract trajectory features and context features automatically. Offline training and online recognition are adopted to complete the target recognition task. Experimental analysis and verification are carried out using the automatic identification system (AIS) dataset. The recognition accuracy increases by 7.91% after context enhancement, which shows that the context enhancement is efficient. The proposed method also has a strong anti-noise ability. In the noisy environment set in this paper, the recognition accuracy of the proposed method is still maintained at 86.13%.

Based on the transformer model, a deep data association and track filtering network (DeepAF) was constructed in this paper to achieve the function of data association and end-to-end track filtering. Combined with the existing track initiation methods, DeepAF can be used to track multiple targets in clutter environments. Experimental results show that DeepAF can stably and effectively track targets moving in different models such as constant velocity, constant acceleration, and constant turn rate. Compared with the probability hypothesis density filter and the probabilistic data association method, which were set with different state transition matrices manually to match with the actual target motion models, DeepAF has similar estimation accuracy in respect of target velocity and better estimation accuracy in respect of target position with less time consumption. For position estimation, compared with PHD, DeepAF can reduce the estimation error by 49.978, 49.263, and 2.706 m in the CV, CA, and CT motion models. Compared with PDA, DeepAF can reduce the estimation error by 13.465, 23.98, and 4.716 m in CV, CA, and CT motion models. For time consumption, compared with PHD, DeepAF can reduce the time by 991.2, 982.3, and 979.5 s in CV, CA, and CT motion models. Compared with PDA, DeepAF can reduce the time by 61.6, 60.5, and 61.4 s in CV, CA, and CT motion models.