In array signal processing, the direction of arrival (DOA) of the signal source has drawn broad research interests with its wide applications in fields such as sonar, radar, communications, medical detection, and electronic countermeasures. In recent years, the application of deep learning (DL) to DOA estimation has achieved great success. This study provides a systematic review of research on DOA estimation using deep neural network methods. We manually selected twenty-five papers related to this research from five prominent databases (SpringerLink, IEEE Xplore, ScienceDirect, Scopus, and Google Scholar) for exploration. Six questions describing the overall trend of DOA estimation using deep learning are put forward. Then, we answered these questions by reviewing the literature. This review is helpful for researchers in this field because it provides more specific and comprehensive information needed for future research. Specifically, we first analyzed the background of the selected papers, including the type of publication, the number of citations, and the country of origin. Then, the DL technology used in DOA estimation is systematically analyzed, including the purpose of using DL in DOA estimation, various DL models (convolutional neural network, deep neural network, and combination network), and various DOA estimation schemes. Finally, various evaluation criteria (root-mean-squared error, accuracy, and mean absolute error) are used to evaluate the DL technology in DOA estimation, and various factors (signal-to-noise ratio, number of snapshots, number of antennas, and number of signal sources) affecting DOA estimation are analyzed. Based on our findings, we believe that deep learning can perform DOA estimation well, and there is still room for improvement in deep learning technology. In this study, the factors affecting DOA estimation can be used as the direction for researchers to conduct in-depth research.

In signal processing, empirical mode decomposition (EMD) first decomposes the received single-channel signal into several intrinsic mode functions (IMFs) and a residual, and then uses machine learning methods for source number enumeration. EMD, however, has an end effect that can undermine the accuracy of source number enumeration. To address this issue, this paper proposed a new EMD method named Supplementary Empirical Mode Decomposition (SEMD), which improved the accuracy by extending the signal length. The proposed method can be better applied to the modal parameter identification of non-stationary and nonlinear data in the engineering field. This method first identifies two candidate extreme points, which are the closest to the function value of the first extreme point near the endpoint. Then, on one side of the candidate point, it finds a waveform similar to that at the endpoint. Finally, the maximum and minimum points at each end of the signal will be added to extend the length of the signal. The added extreme points are candidate extreme points in similar waveforms. For the improved source number enumeration method based on SEMD, the instantaneous phase is obtained first by SEMD and Hilbert transform (HT). Then, the instantaneous phase feature is extracted to obtain a high-dimensional eigenvalue vector. Finally, the back propagation (BP) neural network is used to predict the number of sources. Experiment shows that SEMD can effectively restrain the end effect, and the source number enumeration algorithm based on SEMD has a higher correct detection probability than others.

Source number estimation is a key challenge in multi-sensor array signal processing, focused on accurately determining the number of signal sources based on observed data. This problem is vital for applications in radar, sonar, wireless communication, and astronomy. Despite the variety of methods devel-oped for source number estimation, a comprehensive systematic literature review (SLR) is lacking. This re-view begins by outlining traditional methods for source number estimation and assessing their performance through simulations. It then delves into the latest advancements made between 2017 and 2025, aiming to provide a thorough overview of the factors influencing source number estimation. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, seven digital databases were searched, and 23 key papers were selected after applying predefined inclusion and exclusion criteria. The results show that the performance of source number estimation algorithms is heavily impacted by varia-bles such as noise background, signal-to-noise ratio (SNR), the number of snapshots, array manifold, and array element number. Finally, this review points to future research opportunities, particularly the potential of deep learning techniques, and discusses unresolved challenges and gaps in current studies, offering rec-ommendations to steer future research in this area.

The human body generates infrared radiation through the thermal movement of molecules. Based on this phenomenon, infrared images of the human body are often used for monitoring and tracking. Among them, key point location on infrared images of the human body is an important technology in medical infrared image processing. However, the fuzzy edges, poor detail resolution, and uneven brightness distribution of the infrared image of the human body cause great difficulties in positioning. Therefore, how to improve the positioning accuracy of key points in human infrared images has become the main research direction. In this study, a multi-scale convolution fusion deep residual network (Mscf-ResNet) model is proposed for human body infrared image positioning. This model is based on the traditional ResNet, changing the single-scale convolution to multi-scale and fusing the information of different receptive fields, so that the extracted features are more abundant and the degradation problem, caused by the excessively deep network, is avoided. The experiments show that our proposed method has higher key point positioning accuracy than other methods. At the same time, because the network structure of this paper is too deep, there are too many parameters and a large volume of calculations. Therefore, a more lightweight network model is the direction of future research.

Few-shot image classification (FSIC) is a critical task in computer vision that aims to accurately classify new categories with only a limited number of labeled examples. This capability is essential for practical applications such as medical diagnosis, remote sensing, and autonomous driving, where data scarcity is common. Although numerous FSIC methods have been proposed, a comprehensive systematic literature review (SLR) remains absent. This review begins by summarizing foundational FSIC approaches and evaluating their effectiveness across standard benchmarks. It then explores recent advancements from 2019 to 2025, providing an in-depth analysis of key factors affecting FSIC performance, including dataset characteristics, model architecture, and evaluation protocols. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, multiple academic databases were searched, resulting in the selection of 26 relevant studies based on strict inclusion and exclusion criteria. The findings highlight that FSIC performance is strongly influenced by factors such as the number of support samples, domain shifts, label noise, and model complexity. Finally, the review identifies future research directions, emphasizing the integration of domain adaptation, lightweight model design, and multimodal learning, while addressing current challenges to guide the next phase of FSIC development.

Empirical Mode Decomposition (EMD) is a widely adopted adaptive signal analysis method, particularly effective for source number estimation in non-stationary environments. It decomposes complex signals into a series of Intrinsic Mode Functions (IMFs) and a residual, thereby facilitating the capture of localized time-frequency characteristics. However, standard EMD is prone to limitations such as mode mixing and end effects, which degrade decomposition fidelity and impair estimation accuracy. To overcome these shortcomings, this study proposes Supplementary Enhanced Empirical Mode Decomposition (SEEMD), an improved variant of EMD designed to enhance decomposition robustness. By injecting controlled white noise and extending signal boundaries, SEEMD significantly mitigates end effects and mode mixing, leading to more reliable IMF extraction. Following decomposition, the Hilbert Transform (HT) is applied to derive the instantaneous phase of each IMF, which serves as the basis for feature extraction. These features are subsequently fed into an ensemble learning framework for robust source number estimation. Experimental results verify that the proposed SEEMD-based approach, when integrated with ensemble learning, out-performs conventional methods, especially under low signal-to-noise ratio (SNR) and limited snapshot conditions. Moreover, the method demonstrates strong adaptability and resilience in real-world signal environments.