The proceedings contain 60 papers. The topics discussed include: comparison of adaptive array-processing schemes for land mine detection using hyperspectral imagery;a novel 2-D DOA estimation algorithm with superior r...
详细信息
The proceedings contain 60 papers. The topics discussed include: comparison of adaptive array-processing schemes for land mine detection using hyperspectral imagery;a novel 2-D DOA estimation algorithm with superior resolution and reduced sidelobes;use of spectral anomaly signature feature to improve ROC performance;an optimization-based parallel particle filter for multitarget tracking;frame-building algorithm for electronically scanned array radar;comparison of parametric and non-parametric feature-aided association;multiple target tracking using Janossy measure density functions;Poisson models for extended target and group tracking;and distributed communications resource management for tracking and surveillance networks.
It is well-known that classical direction of arrival (DOA) estimation methods work well in the case of large samples. However, these methods may be theoretically invalid in the case of small samples, which frequently ...
详细信息
It is well-known that classical direction of arrival (DOA) estimation methods work well in the case of large samples. However, these methods may be theoretically invalid in the case of small samples, which frequently occur in large array systems. Such a large array has two effects: i) The number of samples is relatively quite small, and ii) the dimension of samples is very large. To handle the above problems, a more appropriate method for solving DOA estimators in the case of high-dimensional and small samples is proposed in this paper. First, considering the special structure of received samples, an alternative well-estimated second-order statistic, known as the Gram matrix, is originally constructed to better utilize the spatial and statistical information of signals and noise contained by small samples. Second, two novel methods for estimating the number of targets are derived by combining the Gram matrix and information-theoretic criteria. Third, a novel object function and the corresponding algorithm based on the Gram matrix are designed to estimate the signal subspace more efficiently, and then the DOAs of targets are obtained by multiple signal classification methods. In particular, the theoretical analysis indicates that the improved signal subspace estimation algorithm only needs to decompose the low-dimensional Gram matrix instead of the high-dimensional sample covariance matrix. Finally, simulation results are provided to demonstrate the high accuracy and lower computational complexity of the proposed methods in the case of high-dimensional and small samples.
Detection of multiple closely spaced targets with range-Doppler (RD) migration is a challenging issue for radars, because range cell migration (RCM) and Doppler frequency migration (DFM) during the coherent processing...
详细信息
Detection of multiple closely spaced targets with range-Doppler (RD) migration is a challenging issue for radars, because range cell migration (RCM) and Doppler frequency migration (DFM) during the coherent processing interval (CPI), as well as high sidelobes of strong targets, may deteriorate the performance of target detection and parameter estimation. To realize migration correction and sidelobe suppression simultaneously, a joint iterative adaptive approach (IAA) based on RD processing outputs (RD-JIAA) is first proposed in this article. The input data of RD-JIAA are selected within a smallprocessing window centered around the response peak trajectory in range-velocity domain obtained by the RD processing. Compared with IAA and wideband IAA (WIAA), RD-JIAA has low computational burden. Some instructive suggestions on the selection of processing window sizes are presented considering that most of the target energy should be included in the processing window. Then, a fast implementation, namely, RD-JIAA based on the signal sparsity (RD-SJIAA), is presented to further improve the computational efficiency with tolerable performance loss. Both RD-JIAA and RD-SJIAA are able to utilize the structure relationships between covariance matrices of adjacent range cells to reduce the computational complexity. Finally, the performance of the proposed methods is evaluated by numerical examples.
Drones are widely used in fields such as agriculture, environmental protection, and public safety. In these applications, the ability to detect smalltargets typically directly determines the effectiveness of drone im...
详细信息
Drones are widely used in fields such as agriculture, environmental protection, and public safety. In these applications, the ability to detect smalltargets typically directly determines the effectiveness of drone image analysis. Due to the small number of pixels in the image, feature extraction is very difficult for smalltargets. Traditional algorithms find it difficult to capture the details of smalltargets. Although multi-scale feature fusion technology can improve detection capability, feature loss and interference still occur after multiple samplings. To effectively address this challenge, an innovative architecture called Auxiliary Reversible Bidirectional Feature Pyramid Network (ARBFPN) has been proposed. The core design concept is to enhance the integrity of feature information by introducing auxiliary structures, and to prevent feature loss during transmission by using residual connections, thereby preserving more detailed information, which is crucial for small object detection in the feature extraction stage. Meanwhile, by optimizing the detection head through detail enhancement mechanism and gating mechanism, a Lightweight Detail Enhanced Gated Head (LDEGH) was innovatively proposed to improve the overall detection accuracy. To verify the effectiveness of the proposed architecture, relevant experiments were conducted on the VisDrone2019 dataset. The experimental results show that compared with existing technologies, its performance is significantly better than the state-of-the-art technology (SOTA), bringing new breakthroughs to the field of small object detection in drone images.
Existing infrared small target detection (IRSTD) methods mainly rely on the assumption that the training and testing data come from the same distribution, a premise that does not hold in many real-world scenarios. Add...
详细信息
Existing infrared small target detection (IRSTD) methods mainly rely on the assumption that the training and testing data come from the same distribution, a premise that does not hold in many real-world scenarios. Additionally, the inability to access source domain data in numerous IRSTD tasks further complicates the domain adaptation process. To address these challenges, we propose a novel Source-Free Domain Adaptation (SFDA) framework for IRSTD, denoted as IRSTD-SFDA. This framework comprises two key components: Multi-expert Domain Adaptation (MDA) and Multi-scale Focused Learning (MFL). MDA leverages the source model to generate pseudo masks for the target domain, facilitating the transfer of knowledge from the source to the target domain. To account for the inherent diversity of smalltargets across domains, MDA refines these pseudo masks through a series of operations, including target localization, rolling guidance filtering, shape adaptation, and multi-expert decision, thereby mitigating morphological discrepancies between the source and target domains. Meanwhile, MFL employs a global-local fusion strategy to focus on critical regions, enhancing the model's ability to detect small infrared targets. Extensive experimental evaluations across various cross-domain scenarios demonstrate the effectiveness of the proposed framework.
DOA estimation is one of the core tasks in radar signalprocessing and of great significance in communication, military, and other daily activities. The traditional DOA estimation algorithms has a high time cost for a...
详细信息
DOA estimation is one of the core tasks in radar signalprocessing and of great significance in communication, military, and other daily activities. The traditional DOA estimation algorithms has a high time cost for angle estimation, while most available data-driven estimation methods cannot achieve high-precision estimation since converting DOA estimation into multi-label classification task. Based on this, this paper has proposed a dual layer DOA estimation model based on Support Vector Machine(SVM) classification and Random Forest (RF) regression (SVM-RF) for exactly and rapidly estimating the angle of signal sources received from the Radar antenna array system, which transforms the DOA estimation problem into regression task. This method extracts triangular data on the signal covariance matrix as feature input, first divides the dataset into several small sample sets using SVM, and then it conducts RF regression training separately. Six experiments were conducted to verify the validity, reliability and robustness in different SNRs and snapshots for single and multi-targets. Simulation results indicate that the SVM-RF proposed in this paper has implemented the super-resolution DOA Estimation with the estimation accuracy comparable to the classical algorithm MUSIC and CNN. Moreover, SVM-RF has low computational cost and can perform real-time DOA estimation.
暂无评论