The rapid expansion of satellite constellations and the increasing complexity of satellite communication systems have made space situational awareness crucial for ensuring space security and supporting space activities. Among the various communication systems, satellite telemetry signals, particularly those using composite modulation, play a critical role in tracking and managing satellite data. However, recognizing composite modulated signals remains a significant challenge due to the dynamic nature of channel conditions, low signal-to-noise ratios, and the inherent complexity of these signals. In this paper, we propose a novel framework, CLW, for satellite telemetry signal recognition. The CLW framework utilizes a deep approximate representation module combined with adaptive feature aggregation to address these challenges. Specifically, it extracts multiresolution features at different frequency levels from composite modulated telemetry signals, leveraging a channel-space attention mechanism to efficiently aggregate these features. This approach enhances the feature representation quality, reduces the impact of noise and interference, and improves signal recognition accuracy. Extensive simulations and experiments based on high-orbit satellite signal characteristics validate the performance of the proposed CLW framework. The results demonstrate its robustness in handling challenging communication environments and its superior ability to recognize satellite signals compared to existing methods.

The rapid proliferation of low Earth orbit (LEO) satellite constellations is gradually reshaping the global communication landscape. In the future, satellite communications will become an integral part of integrated ground-air-space networks, but this also introduces significant security challenges, such as spoofing and unauthorized access. Radio Frequency Fingerprinting (RFF), leveraging hardware-induced signal characteristics, offers a promising solution for physical-layer authentication. However, the highly dynamic LEO environment-characterized by rapid relative motion, Doppler effects, and volatile propagation conditions-makes extracting robust RF fingerprints complex. In this study, we propose a novel RFF authentication scheme tailored for LEO satellites within the Iridium constellation. By leveraging deep learning techniques, we extract and classify RF fingerprints directly from raw IQ samples, avoiding extensive preprocessing or demodulation. Our system integrates a Doppler-based signal-matching method to ensure accurate labeling and employs a Multi-Scale Inline Star Convolutional Network (MSLS) to enhance feature extraction. Experimental results demonstrate that our method achieves an authentication accuracy of 88% using significantly fewer training samples compared to existing approaches, making it more efficient and practical for real-world deployment. This work paves the way for robust, scalable, low-cost authentication solutions for satellite communication networks.

This letter proposes a novel multiple extended target (ET) joint tracking and classification (JTC) algorithm based on Gaussian processes (GPs) and labeled multi-Bernoulli (LMB) filter, called the ET-JTC-GP-LMB filter, which aims to track and classify simultaneously multiple ETs with the goal of improving estimation performance. Firstly, we construct the relationship between GP-based extension state and prior class information (PCI), and design a new class probability update method. Then, we integrate these two works into the GP-based ET LMB filtering framework, propose the ET-JTC-GP-LMB filter, and provide its gamma-Gaussian-Gaussian mixture implementation to form a closed recursion. Finally, we present an evaluation metric called class recognition rate (CRR) to evaluate classification performance. The simulation results demonstrate the superior performance of the proposed filter.

Few-shot learning aims to recognize new objects using a few labeled samples. Metric-based methods can effectively achieve this goal by building an appropriate feature space, for which a convolutional neural network (CNN) often acts as the backbone network. Unfortunately, CNN is adept at processing local spatial or channel neighborhood information, resulting in a lack of images’ global representation. In this paper, a global information embedding network (GIEN) is built for few-shot learning tasks. We have made two improvements in GIEN: one is the global information embedding (GIE) block, which captures the long-range dependencies of the images by a learnable frequency filter; the other is the two-branch training strategy, which combines cross-entropy loss and supervised contrastive loss to further promote discriminative features. Equipped with GIE, the proposed two-branch framework can obtain a stronger feature representation and achieve competitive performance on few-shot classification benchmark datasets such asminiImageNet andtieredImageNet.

Object detection has developed rapidly with the help of deep learning technologies recent years. However, object detection on drone view remains challenging due to two main reasons: (1) It is difficult to detect small-scale objects lacking detailed information. (2) The diversity of camera angles of drones brings dramatic differences in object scale. Although feature pyramid network (FPN) alleviates the problem caused by scale difference to some extent, it also retains some worthless features, which wastes resources and slows down the speed. In this work, we propose a novel High-Resolution Feature Pyramid Network (HR-FPN) to improve the detection accuracy of small-scale objects and avoid feature redundancy. The key components of HR-FPN include a high-resolution feature alignment module (HRFA), a high-resolution feature fusion module (HRFF) and a multi-scale decoupled head (MSDH). HRFA feeds multi-scale features from backbone into parallel resampling channels to obtain high-resolution features at the same scale. HRFF establishes a bottom-up path to distribute context-rich low-level semantic information to all layers that are then aggregated into classification feature and localization feature. MSDH cope with the scale difference of objects by predicting the categories and locations corresponding to different scales of objects separately. Moreover, we train model by scale-weighted loss to focus more on small-scale objects. Extensive experiments and comprehensive evaluations demonstrate the effectiveness and advancement of our approach.

LiDAR, as an excellent sensor, can provide positions, motion states, and other objective attribute information of objects in the 3D world. Inevitably, the inherent sparsity of point cloud and the problem of occlusion tend to cause incomplete semantic and geometry information of long-range small objects, posing challenges to 3D object detection. The multi-view models take advantage of the complementary information among bird’s eye view (BEV), range view (RV), and other views to alleviate the above issues. However, most of the existing methods coarsely learn the views’ features and neglect the learning of semantic information, which further leads to unsatisfactory detection performance. To this end, this paper proposes a Local-to-Global Semantic Learning Network (LGSLNet) for multi-view 3D object detection from point cloud. The proposed LGSLNet can effectively learn semantic information to explore the local semantics contained in various channels of RV features and to fuse them with BEV features. It has two branches with different backbones. In the BEV branch, the voxels quantized from the point cloud are extracted by sparse convolutional networks and compressed to BEV features. In the RV branch, a multi-scale backbone with semantic-aware convolution (SAC) is designed to learn the local semantic information of the RV. It allows for adaptation to the 3D location using the auxiliary network. In the fusion module, the bidirectional cross-view channel attention (Bi-CCA) is designed to compensate for the semantic information between multiple views and aggregate new RV and BEV features. Extensive experiments on the KITTI, ONCE, and nuScenes 3D object detection datasets demonstrate the superiority of our proposed method.

Video object detection remains a challenging task due to appearance degradation in certain frames. Existing studies usually aggregate temporal information from multiple frames to enhance the object’s appearance representation. Although significant detection performance has been achieved, there are still two shortcomings:(1)The spatial context information within each frame is not fully exploited, which can provide additional decision support when objects are corrupted;(2)In the feature alignment phase, traditional methods tend to employ one-to-one or one-to-global temporal alignment strategies, overlooking the local temporal correlation of objects. To address the above issues, we propose a Joint Spatial and Temporal Feature Enhancement Network (JSTFE-Net) for video object detection, which can jointly utilize spatial-temporal information. First, we present a novel local-global context enhancement module to effectively encode intra-frame spatial context information. This module can enhance the learning of both local details and global semantic information of objects, thereby facilitating accurate object perception within the spatial domain. Second, we develop a deformable temporal sampling module, which adaptively samples correlated temporal information according to the motion information between frames. In addition, to improve the aggregation of temporal-correlated sampled features from multiple frames, we devise an attention-based temporal aggregation block, which dynamically fuses these feature points based on their temporal similarity with the corresponding object feature point. Note that our JSTFE-Net can be effortlessly plugged into image object detectors and state-of-the-art video object detectors. Extensive experiments on the ImageNet VID dataset show that the proposed JSTFE-Net can consistently and significantly improve performance, demonstrating its effectiveness in video object detection.

Graph convolutional neural network (GCN) has drawn increasing attention and attained good performance in various computer vision tasks, however, there is a lack of a clear interpretation of GCN’s inner mechanism. For standard convolutional neural networks (CNNs), class activation mapping (CAM) methods are commonly used to visualize the connection between CNN’s decision and image region by generating a heatmap. Nonetheless, such heatmap usually exhibits semantic-chaos when these CAMs are applied to GCN directly. In this paper, we proposed a novel visualization method particularly applicable to GCN, Vertex Semantic Class Activation Mapping (VS-CAM). VS-CAM includes two independent pipelines to produce a set of semantic-probe maps and a semantic-base map, respectively. Semantic-probe maps are used to detect the semantic information from the semantic-base map to aggregate a semantic-aware heatmap. Qualitative results show that VS-CAM can obtain heatmaps where the highlighted regions match the objects much more precisely than CNN-based CAM. The quantitative evaluation further demonstrates the superiority of VS-CAM.

Tracking-by-detection algorithms have considerably enhanced tracking performance with the introduction of recent convolutional neural networks (CNNs). However, most trackers directly exploit standard scalar-output CNN features, which may not capture enough feature encoding information, instead of aggregated CNN features of vector-output form. In this paper, we propose an end-to-end feature aggregation capsule framework. First, based on the existing CNN network, we aggregate a certain number of similar position-aware CNN features into a capsule to model the feature similarity. The acquired vector-level feature capsules (rather than previous scalar-level pointwise features) are utilized for differentiation learning. We then propose a group attention module to better model the entity representation between different capsule groups thus optimizes total discriminative capability. Third, to reduce the prediction interference resulted by the side effect of dimension rising within capsules, we propose a penalty attention module. Such strategy could dynamically adjust values of neurons by estimating whether they are beneficial or harmful to tracking. Experimental results on five representative benchmarks (UAVDT, DTB70, UAV123, VOT2016 and VOT2018) demonstrate the excellent tracking performance of our dual attention based capsule tracker (DACapT). Specially, it exceeds the previous top tracker by 4.6%/1.9% in precision/success evaluations on UAVDT.

Cyclostationary signals analysis is a highly widespread tool for non-stationary signals processing. In practical transmission, signals may be contaminated by non-stationary non-Gaussian impulse noises. However, cyclostationary is weak in dealing with such non-Gaussian distributions. In this paper, a cyclic mean kernel function is proposed for cyclostationary signal analysis under impulse noise, based on a high-dimensional space transformation. Then, the generalized cyclic mean kernel function is proposed based on the constructed generalized Hankel matrix. The simulation results show that the cyclostationary signal frequency estimation methods based on these two functions have advantages in performance and robustness over the existing methods under impulsive noise.

In this work, the authors propose a novel self‐supervised learning method based on mutual information to learn representations from the videos without manual annotation. Different video clips sampled from the same video usually have coherence in the temporal domain. To guide the network to learn such temporal coherence, they maximise the mutual information between global features extracted from different clips sampled from the same video (Global‐MI). However, maximising the Global‐MI leads the network to seek shared content from different video clips and may make the network degenerate to focus on the background of the video. Considering the structure of the video, they further maximise the average mutual information between the global feature and local patches of multiple regions of the video clip (multi‐region Local‐MI). Their approach, which is called Max‐GL, learns the temporal coherence by jointly maximising the Global‐MI and multi‐region Local‐MI. Experiments are conducted to validate the effectiveness of the proposed Max‐GL. Experimental results show that the Max‐GL can serve as an effective pre‐training method for the task of action recognition in videos. Additional experiments for the task of action similarity labelling and dynamic scene recognition also validate the generalisation of the learned representations of the Max‐GL.

Human action recognition in videos is an important research topic in computer vision due to its wide applications. Actions naturally contain both spatial and temporal information. The key to action recognition is to model the spatial and temporal structures of actions. In this study, the authors propose an attention‐based spatial–temporal hierarchical convolutional long short‐term memory (ST‐HConvLSTM) network to model the structures of actions in the spatial and temporal domains. The ST‐HConvLSTM consists of two parts: a spatial–temporal attention module and a novel LSTM‐like architecture named hierarchical ConvLSTM (HConvLSTM). The HConvLSTM can model the spatial and temporal structures of actions. The spatial–temporal attention module can figure out which part of the video is more discriminative for action recognition and makes the HConvLSTM focus on it. In addition, a weighted fusion strategy is proposed to fuse the appearance information and motion information of the video. The proposed ST‐HConvLSTM is evaluated on UCF101, HMDB51 and Kinetics datasets. Experimental results show that the authors’ proposed ST‐HConvLSTM achieves state‐of‐the‐art performance compared with other recent LSTM‐like architectures and attention‐based methods.

The level set method based on bias correction can segment images with gentle intensity inhomogeneity effectively. However, most level set methods fail to segment severe inhomogeneous images due to the use of fixed scale clustering criterion. To deal with this problem, an adaptive multilayer level set method is proposed to segment images with severe intensity inhomogeneity. First, an improved global adaptive scale operator and a local adaptive scale operator are designed to adaptively adjust the scale of clustering kernel function according to the degree of intensity inhomogeneity. Then, an adaptive multilayer level set structure is constructed with the two designed scale operators. The number of layers and the scale of each layer in the multilayer structure are adaptively determined based on the degree of intensity inhomogeneity, which not only provides appropriate candidate scales in each pixel but also allows the model to detect global contrast information. With the dual minimisation method, image segmentation and bias correction can be achieved simultaneously. In addition, a hybrid bias field initialisation procedure is proposed to enhance the robustness of the proposed method. Experimental results demonstrate the effectiveness and robustness of the proposed method in segmenting images with intensity inhomogeneity.

This study presents an improved multi‐target multi‐Bernoulli (IMeMBer) gamma Gaussian inverse Wishart (GGIW) filter for tracking multiple extended targets (ETs). The main contribution of this study consists of three parts, first, a novel method is proposed to obtain the unbiased cardinality estimation of multiple targets using the multi‐Bernoulli recursion. As a variation of the existing cardinality‐balanced MeMBer (CBMeMBer) filter, the presented filter is called the improved MeMBer filter, which overcomes the high detection probability limitation of the CBMeMBer filter. Second, based on the mathematical derivation, the IMeMBer filter is expanded to accommodate the characteristics of the ETs of which each target generates more than one measurement at each time step, and the GGIW method is used for its implementation. The resulting filter simultaneously provides the kinematic, extended and measurement rate states of ETs with an unknown and time‐varying number. Third, the simulation results show that the presented filter achieves a considerable performance at the cost of less time, compared to the labelled multi‐Bernoulli GGIW filter.

Specific emitter identification (SEI) is gaining popularity since it can distinguish different individuals in same type of radar emitter under complex electromagnetic environment. However, classification of signals is still a challenging task when the feature has low physical representation. In this work, we propose a compressed sensing mask feature in ambiguity domain, which can significantly improve the recognition rate of civil flight radar emitters. Furthermore, it not only represents physical characteristics of measured radar signals but also contains more time varying information and alleviates the computational costs. The physical significance and effectiveness of the proposed feature can be verified by reconstructing Wigner-Ville distribution (WVD) from the sparsest ambiguity function. Experimental results corroborate the highly accuracy and stability of the proposed approach.

Appropriately modelling target‐birth intensity is a significant but challenging issue in multi‐target tracking systems. In existing cardinality‐balanced multi‐target multi‐Bernoulli (CBMeMBer) filters, a priori knowledge about the locations where targets appear is required to model the target‐birth intensity. Since the newborn targets can appear anywhere within the observation region, it is impractical to obtain such prior information. In this study, a novel CBMeMBer filter with adaptive target‐birth intensity is presented, considering the newborn and surviving targets separately. The target‐birth function of the target‐birth intensity is modelled using current measurements rather than the known birth locations, and the target‐birth magnitude is assigned by an allocation function rather than equally assigned. The new CBMeMBer filter can remove the restriction on the requirement of prior birth location information and can adapt well after continuous missing detection occurs. Simulations of the sequential Monte Carlo and Gaussian mixture implementations demonstrate the effectiveness of the proposed filter.

The anisotropy of radar cross section of scatterers makes the scatterer trajectory matrix incomplete in sequential inverse synthetic aperture radar images. As a result, factorization methods cannot be directly applied to reconstruct the 3-D geometry of scatterers without additional consideration. We propose a modified expectation-maximization (EM) algorithm to retrieve the complete scatterer trajectory matrix. First, we derive the motion dynamics of the projected scatterer, which approximates an ellipse. Then, based on the estimated ellipse parameters using the known data of each scatterer trajectory, we use the Kalman filter to initialize the missing data. To address the limitations of a traditional EM, which only considers the rank-deficient characteristics of the scatterer trajectory matrix, we propose to augment EM by using both the known rank-deficient and elliptical motion characteristics. Experimental results on simulated data verify the effectiveness of the proposed method.

In real world multiple extended target tracking problems, the presence of merged measurements is a frequently occurring phenomenon, however, most existing tracking algorithms in the literature assume that each target generates independent measurements. When this measurement merging phenomenon occurs, it increases the computational complexity of the tracking algorithms. Recently, the conditional joint decision and estimation (CJDE) algorithm based on the generalized Bayes risk was proposed to solve problems of joint detection, tracking and classification (JDTC) of targets. In this paper, we develop a principled Bayesian solution to the important problem involving inter-dependent decision and estimation conditioned on data based on the theory of random finite sets (RFS), and a tractable implementation based on the recently proposed generalized labeled multi-Bernoulli (GLMB) filter. The performance of the proposed technique is demonstrated by simulation of a multi-target bearings-only tracking scenario, where measurements become merged due to finite resolution effects.

The Gaussian inverse Wishart (GIW) filter is a promising filter for extended target tracking and draws tremendous attention in recent years. The Gaussian and the inverse Wishart distributions are used to describe the target's kinematical and extended states, respectively. However, the filter for estimating the extended state contains predicting position error and causes large error of the extended state estimation, especially for the scenarios with high-maneuvering. In this paper, we eliminate the influence of the predicting position error via reconstructing the updated equation for estimating extended state. Based on GIW probability hypotheses density (GIW-PHD) framework, the improved filter is tested in a maneuvering scenario and the comparative results verify the superior performance of the filter in terms of the extended state estimation.

This paper presents an approach named symmetric measurement equation (SME) to track known number of multiple extended targets. The SME approach removes the target-measurement association uncertainty through converting the original observation into pseudo-measurement vector. The work is focused on tracking moving extended target using SMEs which define new measurements through the sums of products of the original measurements. The performance of the SME filter for multiple extended targets tracking is demonstrated by a computer simulation compared with the ETT-PHD filter.

The random-matrix approach to extended object tracking (EOT) supposes that the measurement model is linear and its covariance is a random matrix to stand for the object extension. In practice, however, most measurement algorithms are nonlinear and multiple extensions cannot be simplified by an ellipsoid. This paper proposes a new method for nonlinear maneuvering non-ellipsoidal extended object tracking (non-MNEOT) based on the two problems above in practice. Firstly, the stochastic model approximation represented by matched linearization is used to solve the nonlinear measurement model and the linearized measurements are applied to the random-matrix algorithms to the EOT after a simple conversion. Then, multiple sub-objects are used to approximate the non-ellipsoidal target and applied to the extended object tracking by a combination of random matrix. Lastly, most targets in practice are maneuvering targets, so we introduce the interactive multiple model (IMM) to nonlinear non-ellipse EOT. The effectiveness of the approach in this paper is demonstrated by simulation results.

The probability hypothesis density (PHD) filter is a promising filter for multi-target tracking which propagates the posterior intensity of the multi-target state. In this paper, a Gaussian mixture particle flow PHD (GMPF-PHD) filter is proposed which uses a bank of particles to represent the Gaussian components in the Gaussian mixture PHD (GM-PHD) filter. Then a particle flow is implemented to migrate the particles to a more appropriate region in order to obtain a more accurate approximation of the posterior intensity. To verify the effectiveness of the algorithm, both linear and nonlinear multi-target tracking problem are designed, and the performance are compared with the classical approaches such as the GM-PHD filter, the Gaussian mixture particle PHD (GMP-PHD) filter, and the particle PHD filter. Simulation results show that the proposed filter can achieve a good performance with a reasonable computational cost.

Extended target Gaussian inverse Wishart PHD filter is a promising filter. However, when the two or more different sized extended targets are spatially close, the simulation results conducted by Granström et al . show that the cardinality estimate is much smaller than the true value for the separating tracks. In this study, the present authors call this phenomenon as the cardinality underestimation problem, which can be solved via a novel robust clustering algorithm, called Bayesian partition, derived by combining the fuzzy adaptive resonance theory with Bayesian theorem. In Bayesian partition, alternative partitions of the measurement set are generated by the different vigilance parameters. Simulation results show that the proposed partitioning method has better tracking performance than that presented by Granström et al., implying good application prospects.

Compared to globality based supervised dimensionality reduction methods such as Fisher Discriminant Analysis (FDA), locality based ones including Local Fisher Discriminant Analysis (LFDA) have attracted increasing interests since they aim to preserve the intrinsic data structures and are able to handle multimodally distributed data. However, both FDA and LFDA are usually solved via a ratio trace form to approximate the trace ratio, which is the Fisher's original objective criterion. In this paper, a novel trace optimization framework is presented to solve the original trace ratio problem. It offers an exact solution via mathematical programming and recovers Fisher's maximal separability faithfully. The resulting maximum Local Fisher Discriminant Analysis (maxLFDA) not only inherits the merits of LFDA, but also boosts the classification accuracy in each target subspace with expected maximum trace ratio value. Experiments on a toy example and real-world face databases validate the effectiveness of the proposed method.

Radar emitter recognition is an important and challenging subject in radar signal analysis and processing. In this work, an ambiguity function (AF) representative-slice based feature extraction and optimization algorithm is presented for unintentional modulation recognition of moving radar emitters. It considers near-zero slices of AF as representative feature set of radar emitters, which not only coincides with the characteristics of real radar signals, but also mitigates the computation problem and avoids undesired cross terms in existing AF based method. Direct Discriminant Ratio (DDR) criterion is further utilized to preserve the most discriminant features and boost recognition accuracy, by ranking the kernel points along the representative-slice. Experimental results validate the practical usefulness and high stability of the proposed approach on real data of moving radar emitters, as well as synthetic radar data from U.S. Naval Research Laboratory.

In this paper, we address the problem of entropy-based clustering in the framework of possibility theory. First, we introduce the possibilistic entropy with brief discussion. Then we develop the possibilistic entropy theory for clustering analysis and investigate the general Possibilistic Entropy Clustering (PEC) problems, based on which a Fully Unsupervised Possibilistic Entropy Clustering (FUPEC) algorithm is elaborated in detail with the following advantages: (I) having clearer physical meaning and well-defined mathematical features; (2) automatically determining the number of the clusters; (3) automatically controlling the resolution parameter during the clustering progress; (4) overcoming the sensitivity to initialization and to the noise and outliers. Finally, we illustrate the effectiveness of this novel algorithm with various examples.