Effective feature representations play a decisive role in content-based remote sensing image retrieval (CBRSIR). Recently, learning-based features have been widely used in CBRSIR and they show powerful ability of feature representations. In addition, a significant effort has been made to improve learning-based features from the perspective of the network structure. However, these learning-based features are not sufficiently discriminative for CBRSIR. In this paper, we propose two effective schemes for generating discriminative features for CBRSIR. In the first scheme, the attention mechanism and a new attention module are introduced to the Convolutional Neural Networks (CNNs) structure, causing more attention towards salient features, and the suppression of other features. In the second scheme, a multi-task learning network structure is proposed, to force learning-based features to be more discriminative, with inter-class dispersion and intra-class compaction, through penalizing the distances between the feature representations and their corresponding class centers. Then, a new method for constructing more challenging datasets is first used for remote sensing image retrieval, to better validate our schemes. Extensive experiments on challenging datasets are conducted to evaluate the effectiveness of our two schemes, and the comparison of the results demonstrate that our proposed schemes, especially the fusion of the two schemes, can improve the baseline methods by a significant margin.

The content-based remote sensing image retrieval (CBRSIR) has recently become a hot topic due to its wide applications in analysis of remote sensing data. However, since conventional CBRSIR is unsuitable in harsh environments, this paper focuses on the cross-modality CBRSIR (CM-CBRSIR) between synthetic aperture radar (SAR) and optical images. Besides the large inter-class and small intra-class in CBRSIR, CM-CBRSIR is limited by prominent modality discrepancy caused by different imaging mechanisms. To address this limitation, this study proposes a deep cross-modality hashing network (DCMHN). First, we transform optical images with three channels into four different types of single-channel images to increase diversity of the training modalities. This helps the network to mainly focus on extracting the contour and texture shared features, and make it less sensitive to colour information for images across modalities. Secondly, we combine any type of randomly selected transformed images and its corresponding SAR or optical images to form image pairs that are fed into the networks. The training strategy, with paired image data, eliminates the large cross-modality variations caused by different modalities. Finally, the triplet loss, in combination with the hash function, helps the modal to extract the discriminative features of images and upgrade the retrieval efficiency. To further evaluate the proposed modality, we construct a SAR-optical dual-modality remote sensing image dataset (SODMRSID) containing twelve categories. Experimental results demonstrate the superiority of the proposed method with regards to efficiency and generality.

Currently, most multi-target data association methods require the assumption that the target motion model is known, but this assumption is clearly not valid in a real environment. In the case of an unknown system model, the influence of environmental clutter and sensor detection errors on the association results should be considered, as well as the occurrence of strong target maneuvers and the sudden appearance of new targets during the association process. To address these problems, this paper designs a target tracking and data association algorithm based on reinforcement learning. First, this algorithm combines the dynamic exploration capability of reinforcement learning and the long-time memory function of LSTM network to design a policy network that predicts the probability of associating a point with its various possible source targets. Then, the Bayesian network and the multi-order least squares curve fitting method are combined to predict the location of target, and the results are fed into the Bayesian recursive function to obtain the reward. Simultaneously, some corresponding mechanisms are proposed for possible problems that interfere with the association process. Finally, the simulation experimental results show that this algorithm associates the results with higher accuracy compared to other algorithms when faced with the above problem.

The point‐track association methods are proposed on the premise that the system models are known, which obviously does not conform to the actual air target detection environment. Considering this situation, for the point‐track association problems in clutter environment, a point‐track association method with unknown system model (USMA) is proposed. The method integrates reinforcement learning (RL) theory and traditional point‐track association framework, utilises the association process migration of different models, simplifies the entire learning process, and improves the generalisation ability by designing an adaptive mechanism. The experimental results show that when the system model is unknown, the USMA method can more accurately correlate to the measurements, and can also solve the problems of point‐track association with a certain clutter density. Compared with other methods, the USMA method performs better.

With the continuous expansion of the application field of artificial intelligence, radar data processing has also begun to fully enter the era of intelligence, and new achievements have emerged in the research fields of target detection, target tracking, and target recognition. At a time of rapid development of artificial intelligence technology, it is necessary to think about the future development of radar data intelligent processing. To this end, combining the research history and the superficial understanding of radar data intelligent processing in the past ten years, our team analyzes the main research progress and challenges of radar data intelligent processing, examines the new requirements for autonomous, multi-modal, and multi-mode radar data intelligent processing, and explores new opportunities in the era of large models, etc. Furthermore, along a research path of ``1 domain foundation model + 3 core task models + N typical application implementations'', the research conceptions and main contents of radar data large model are put forward in order to enable the research of intelligent radar.

To address the performance degradation of traditional data association algorithms caused by unknown target motion models, environmental interference, and strong maneuvering behaviors in complex dynamic scenarios, this paper proposes an innovative fusion algorithm that integrates reinforcement learning and deep learning. By constructing a policy network that combines Long Short-Term Memory (LSTM) memory units and reinforcement learning dynamic decision-making, a dynamic prediction model for "measurement-target" association probability is established. Additionally, a hybrid predictor incorporating Bayesian networks and multi-order curve fitting is designed to formulate the reward function. To tackle practical interference, a dynamic ring gate screening mechanism and a trajectory consistency-based error correction module are developed, effectively suppressing clutter interference and enabling autonomous correction of association errors. Experimental results demonstrate that the proposed method significantly improves association accuracy in high-noise environments compared to traditional algorithms, enhancing robustness in complex unknown scenarios.