| 引用本文: | 张转花，靳俊峰，常沛，何洋洋，汪振亚，侯其立，李玉景，郝慧军，曾怡，夏勇，商国军，许涛，任伟杰，雷鸣，王歆远，寿博，邓丽颖，任乐乐，窦曼莉，杨利红，张琦珺，李伟，牛蕾，林晓斌，张志成. BWRadarDataset‑1.0：多波段多模态雷达探测感知数据集[J]. 雷达科学与技术, 2026, 24(1): 1-14.[点击复制] |
| ZHANG Zhuanhua, JIN Junfeng, CHANG Pei, HE Yangyang, WANG Zhenya, HOU Qili, LI Yujing, HAO Huijun, ZENG Yi, XIA Yong, SHANG Guojun, XU Tao, REN Weijie, LEI Ming, WANG Xinyuan, SHOU Bo, DENG Liying, REN Lele, DOU Manli, YANG Lihong, ZHANG Qijun, LI Wei, NIU Lei, LIN Xiaobin, ZHANG Zhicheng. BWRadarDataset‑1.0: Multi‑Band Multi‑Mode Radar Dataset for Detection and Sensing[J]. Radar Science and Technology, 2026, 24(1): 1-14.[点击复制] |
|
| |
|
|
| 本文已被：浏览 502次 下载 76次 |
码上扫一扫！ |
|
|
| BWRadarDataset‑1.0：多波段多模态雷达探测感知数据集 |
|
张转花，靳俊峰，常沛，何洋洋，汪振亚，侯其立，李玉景，郝慧军，曾怡，夏勇，商国军，许涛，任伟杰，雷鸣，王歆远，寿博，邓丽颖，任乐乐，窦曼莉，杨利红，张琦珺，李伟，牛蕾，林晓斌，张志成
|
|
1. 中国电子科技集团公司第三十八研究所， 安徽合肥 230088;2. 雷达探测感知全国重点实验室， 安徽合肥 230088
|
|
| 摘要: |
| 雷达探测感知技术飞速发展浪潮下高质量数据集在算法创新、模型训练与性能验证中发挥着重要作用。当前，深度学习等数据驱动方法已成为提升雷达在检测、跟踪、识别、干扰及合成孔径雷达（SAR）成像等核心任务性能的关键。然而，现有的数据集大多基于仿真生成，与真实电磁环境存在差异，泛化能力受限，并且现有的数据集仅针对单一功能，例仅有检测或SAR，缺乏系统性，难以支撑探测感知处理的一体化研究。针对这一空白，本文公开了一套完整的雷达检测跟踪识别一体化数据集。该数据集源于典型的实测场景，涵盖了信号处理、目标跟踪、精细识别、复合干扰以及高分辨率SAR图像的多波段、多模态数据，真实反映复杂环境下雷达信号的传播特性与目标特性。进一步，本文对数据集中的关键特征进行了系统性提取与分析，为不同任务的算法研究与性能评估提供了标准化的特征输入，为研究雷达智能化信号与信息处理提供了坚实的基础。 |
| 关键词: 雷达探测 公开数据集 特征提取 目标检测 目标跟踪 目标识别 有源干扰 SAR图像 特征分析 |
| DOI：DOI:10.3969/j.issn.1672-2337.2026.01.001 |
| 分类号:TN957 |
| 基金项目:雷达探测感知全国重点实验室基金（2401074240408, KGJ24010742403, KGJ2401072408） |
|
| BWRadarDataset‑1.0: Multi‑Band Multi‑Mode Radar Dataset for Detection and Sensing |
|
ZHANG Zhuanhua, JIN Junfeng, CHANG Pei, HE Yangyang, WANG Zhenya, HOU Qili, LI Yujing, HAO Huijun, ZENG Yi, XIA Yong, SHANG Guojun, XU Tao, REN Weijie, LEI Ming, WANG Xinyuan, SHOU Bo, DENG Liying, REN Lele, DOU Manli, YANG Lihong, ZHANG Qijun, LI Wei, NIU Lei, LIN Xiaobin, ZHANG Zhicheng
|
|
1. The 38th Research Institute of China Electronics Technology Group Corporation, Hefei 230088, China;2. National Key Laboratory of Radar Detection and Sensing, Hefei 230088, China
|
| Abstract: |
| With the rapid development of radar detection and sensing technology, high?quality datasets play a critical role in algorithm innovation, model training, and performance verification. Nowadays, data?driven approaches such as deep learning have become the key in improving radar performance including detection, tracking, recognition, jamming and SAR. However, most of the existing datasets are generated by simulation, which is different from the real electromagnetic environment, and the generalization ability is limited. Moreover, these datasets are often designed for single mode, such as detection or SAR, and lack of systematicness, which is difficult to support the integrated research of detection, sensing and processing. In response to this gap, this paper introduces a comprehensive integrated dataset for radar detection, tracking and recognition. The dataset is derived from typical measured scenes, covering multi?band and multi?mode data, which includes signal processing, target tracking, fine?grained recognition, compound jamming and high?resolution SAR imaging, which truly reflects the propagation characteristics and target characteristics of radar signals in complex environments. Furthermore, this paper conducts a systematic extraction and analysis of key features within the dataset, providing standardized feature input for algorithm development and performance evaluation in different tasks. This work contributes to provide a solid foundation for research in intelligent radar signal and information processing. |
| Key words: radar detection publicly available datasets feature extraction target detection target tracking target recognition active jamming SAR image feature analysis |
|
|
|
|