Wei Yang

Wei Yang

Auto-generated profile, based on publicly available information

Wei Yang

College of Electronic Science and Technology, National University of Defense Technology, Changsha, China

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
28 January 2025 in IEEE Transactions on Image Processing
Verified
Journal Article
Open access
WLWeijie Li
Weijie Li
WYWei Yang
Wei Yang
Yuenan HouYuenan Hou
Yuenan Hou
LLLi Liu
Li Liu
YLYongxiang Liu
Yongxiang Liu
XLXiang Li
Xiang Li
Despite the remarkable progress in synthetic aperture radar automatic target recognition (SAR ATR), recent efforts have concentrated on detecting and classifying a specific category, e.g., vehicles, ships, airplanes, or buildings. One of the fundamental limitations of the top-performing SAR ATR methods is that the learning paradigm is supervised, task-specific, limited-category, closed-world learning, which depends on massive amounts of accurately annotated samples that are expensively labeled by expert SAR analysts and have limited generalization capability and scalability. In this work, we make the first attempt towards building a foundation model for SAR ATR, termed SARATR-X. SARATR-X learns generalizable representations via self-supervised learning (SSL) and provides a cornerstone for label-efficient model adaptation to generic SAR target detection and classification tasks. Specifically, SARATR-X is trained on 0.18 M unlabelled SAR target samples, which are curated by combining contemporary benchmarks and constitute the largest publicly available dataset till now. Considering the characteristics of SAR images, a backbone tailored for SAR ATR is carefully designed, and a two-step SSL method endowed with multi-scale gradient features was applied to ensure the feature diversity and model scalability of SARATR-X. The capabilities of SARATR-X are evaluated on classification under few-shot and robustness settings and detection across various categories and scenes, and impressive performance is achieved, often competitive with or even superior to prior fully supervised, semi-supervised, or self-supervised algorithms. Our SARATR-X and the curated dataset are released at https://github.com/waterdisappear/SARATR-X to foster research into foundation models for SAR image interpretation.

12 October 2023 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
WLWeijie Li
Weijie Li
WYWei Yang
Wei Yang
WZWenpeng Zhang
Wenpeng Zhang
TLTianpeng Liu
Tianpeng Liu
YLYongxiang Liu
Yongxiang Liu
LLLi Liu
Li Liu
Vehicle recognition is a fundamental problem in SAR image interpretation. However, robustly recognizing vehicle targets is a challenging task in SAR due to the large intraclass variations and small interclass variations. Additionally, the lack of large datasets further complicates the task. Inspired by the analysis of target signature variations and deep learning explainability, this paper proposes a novel domain alignment framework named the Hierarchical Disentanglement-Alignment Network (HDANet) to achieve robustness under various operating conditions. Concisely, HDANet integrates feature disentanglement and alignment into a unified framework with three modules: domain data generation, multitask-assisted mask disentanglement, and domain alignment of target features. The first module generates diverse data for alignment, and three simple but effective data augmentation methods are designed to simulate target signature variations. The second module disentangles the target features from background clutter using the multitask-assisted mask to prevent clutter from interfering with subsequent alignment. The third module employs a contrastive loss for domain alignment to extract robust target features from generated diverse data and disentangled features. Lastly, the proposed method demonstrates impressive robustness across nine operating conditions in the MSTAR dataset, and extensive qualitative and quantitative analyses validate the effectiveness of our framework.

28 January 2026 in IEEE Transactions on Pattern Analysis and Machine Intelligence
Verified
Journal Article
Open access
YLYongxiang Liu
Yongxiang Liu
WLWeijie Li
Weijie Li
LLLi Liu
Li Liu
JZJie Zhou
Jie Zhou
PBpeng bowen
peng bowen
XLXiang Li
Xiang Li
The absence of publicly available, large-scale, high-quality datasets for Synthetic Aperture Radar Automatic Target Recognition (SAR ATR) has significantly hindered the application of rapidly advancing deep learning techniques, which hold huge potential to unlock new capabilities in this field. This is primarily because collecting large volumes of diverse target samples from SAR images is prohibitively expensive, largely due to privacy concerns, the characteristics of microwave radar imagery perception, and the need for specialized expertise in data annotation. Throughout the history of SAR ATR research, there have been only a number of small datasets, mainly including targets like ships, airplanes, buildings, etc. There is only one vehicle dataset MSTAR collected in the 1990s, which has been a valuable source for SAR ATR. To fill this gap, this paper introduces a large-scale, new dataset named ATRNet-STAR with 40 different vehicle categories collected under various realistic imaging conditions and scenes. It marks a substantial advancement in dataset scale and diversity, comprising over 190,000 well-annotated samples—$10\times$ larger than its predecessor, the famous MSTAR. Building such a large dataset is a challenging task, and the data collection scheme will be detailed. Second, we illustrate the value of ATRNet-STAR via extensively evaluating the performance of 15 representative methods with 7 different experimental settings on challenging classification and detection benchmarks derived from the dataset. Finally, based on our extensive experiments, we identify valuable insights for SAR ATR and discuss potential future research directions in this field. We hope that the scale, diversity, and benchmark of ATRNet-STAR can significantly facilitate the advancement of SAR ATR.

26 December 2025 in IEEE Aerospace and Electronic Systems Magazine
Verified
Journal Article
Closed access
YLYongxiang Liu
Yongxiang Liu
WYWei Yang
Wei Yang
XQXiangfeng Qiu
Xiangfeng Qiu
CYChen Yang
Chen Yang
WJWeidong Jiang
Weidong Jiang
MGMaria Sabrina Greco
Maria Sabrina Greco
Cognitive radar enhances traditional radar systems by equipping them with the ability to sense, understand, learn, reason, and make decisions, thus enabling these systems to adapt to increasingly complex and dynamic electromagnetic environments. This adaptive capability significantly improves radar performance in specific missions, including target detection, tracking, and recognition. However, despite its promising potential, cognitive radar is still in the early stages of development, transitioning from theoretical concepts to practical implementation, with many foundational theories and technologies remaining underdeveloped. Current works are primarily focused on optimizing waveforms under the assumption of prior knowledge, often overlooking a comprehensive and practical perspective of radar systems. These approaches lead to challenges such as overly idealistic assumptions, difficulties in obtaining prior knowledge, and the complexity of waveform diversity schemes and design algorithms. To advance the practical application of cognitive radar systems, this paper addresses key issues in two critical domains: clutter suppression and jamming countermeasures. We propose solutions based on cognitive radar, focusing on the development of the required hardware systems for these tasks and the design of their closed-loop operation processes. Experimental results, derived from actual measurement data, demonstrate the effectiveness of the proposed architecture in both clutter suppression and jamming countermeasures.

3 July 2025 in IEEE Transactions on Aerospace and Electronic Systems
Verified
Journal Article
Closed access
CYChen Yang
Chen Yang
WYWei Yang
Wei Yang
XQXiangfeng Qiu
Xiangfeng Qiu
WJWeidong Jiang
Weidong Jiang
YLYongxiang Liu
Yongxiang Liu
GFGini Fulvio
Gini Fulvio
Designing a transmit sequence set with desired multidimensional properties can enhance target sensing performance but increases the complexity of problem. In this paper, we focus on designing transmit sequence set in multiple input multiple output radar to obtain desired properties in multiple dimensions. Specifically, a multi-constrained optimization model is formulated, which takes into account the performance in the range-Doppler, spatial, and frequency dimensions. In order to solve the high-order, fractional and non-convex resulting problem, an algorithm based on successive convex approximation (SCA) is proposed, in which each iteration contains semi-definite programming and eigenvalue decomposition. To expedite the iterative convergence, a SCA-acceleration strategy is introduced, which utilizes the square accelerated iteration method in multi-constrained conditions. Finally, numerical experiments validate the superiority of the proposed algorithm over stateof-art methods. Simultaneously, multi-dimensional joint shaping experiments are conducted to investigate the trade-off between the sidelobe performance of the ambiguity function and the beampattern.

9 May 2025 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
Shangqu YanShangqu Yan
Shangqu Yan
YFYaowen Fu
Yaowen Fu
RYRuofeng Yu
Ruofeng Yu
CLChenyang Luo
Chenyang Luo
WZWenpeng Zhang
Wenpeng Zhang
WYWei Yang
Wei Yang
Video synthetic aperture radar (ViSAR) attracts much attention in the field of moving target detection due to its high frame rate and high-resolution imaging capability. In ViSAR, the moving target's shadow does not offset and defocus, which is widely used as a feature for moving target detection. To address the shortcomings of high false and missing alarms in existing detection algorithms, this study proposes a high-precision moving target shadow detection algorithm for ViSAR based on information geometry. Firstly, the shadow enhancement and background suppression algorithm is used to enhance the saliency of moving targets' shadows in ViSAR images to reduce the false and missing alarms caused by complex background. Then, based on the information geometry theory, the enhanced ViSAR image is mapped to a Gaussian statistical manifold, and the Jensen–Shannon (JS) divergence matrix with pixel context information is extracted from this manifold. And the JS divergence matrix and the gray-scale information from the enhanced ViSAR image are used to construct the joint probability distribution function (JPDF). The 2D minimum cross-entropy principle is used to calculate the threshold to extract the regions of interest (ROIs). Finally, the final detection results are obtained by applying single-frame and multiple-frame false alarms suppression. The experimental results, based on real ViSAR dataset published by Sandia National Laboratories (SNL), verify the effectiveness of our proposed algorithm. Specifically, the Recall rate is 85.97%, the Precision is 96.22%, and the F1 score achieves 0.91.