本文针对舰载雷达低、慢、小目标探测面临的小样本识别问题，提出一种基于小样本学习的低慢小目标分类识别方法。该方法将低慢小目标雷达回波数据转换到小波变换域，利用多头注意力机制和双向长短记忆人工神经网络相结合的方式，解决了小样本目标分类识别的问题。在低慢小目标雷达回波仿真数据集上，开展模型训练和算法验证，分析任务差异性与识别准确率的关系，实验结果表明该方法对典型低慢小目标识别精度可达90%以上，在雷达目标识别领域具有较好的应用价值。
Aiming at a small number of sample recognition problems faced by low-slow and small target detection of shipborne radar, a low-slow and small target recognition method with few shot learning based on shipborne radar is proposed. The method converts the radar echo data of low-slow and small targets into the wavelet transform domain, and solves the problem of target recognition by using a combination of multi-headed attention mechanism and bi-directional long short-term memory artificial neural network. Model training and algorithm validation are carried out on the low-slow and small target radar echo simulation dataset, and the relationship between task variability and recognition accuracy is analyzed. Experimental results show that the recognition accuracy of typical low-slow and small targets can reach more than 90%, which verifies the good performance of this method and shows promising application in the field of radar target recognition.
2023,45(18): 123-128 收稿日期：2022-08-29
DOI：10.3404/j.issn.1672-7649.2023.18.021
分类号：TN959.1-7
基金项目：武器装备综合研究项目(2020103280)
作者简介：李轲(1985-),男,博士,教授,研究方向为目标检测与识别技术
参考文献：
[1] 卞伟伟, 邱旭阳, 刘方.“低慢小”目标协同探测综合识别关键技术研究[C]//北京: 第五届中国指挥控制大会, 2017: 264–269.
[2] 李明明, 卞伟伟, 甄亚欣. 国外“低慢小”航空器防控装备发展现状分析[J]. 飞航导弹, 2017, 1(9): 62–70
[3] 向凡夫, 郝冬青, 吴鹏. 针对低慢小目标的雷达信号处理算法[J]. 指挥控制与仿真, 2019, 41(4): 40–46
[4] 苑文楠, 贾彦翔, 吕鑫, 等. “低慢小”目标多体制武器分配优化模型[J]. 电光与控制, 2022, 29(3): 1–5
[5] 许述文, 白晓惠, 郭子薰, 等. 海杂波背景下雷达目标特征检测方法的现状与展望[J]. 雷达学报, 2020, 9(4): 684–714
[6] 晏媛, 孙俊, 孙晶明, 等. 雷达小样本目标识别方法及应用分析[J]. 系统工程与电子技术, 2021, 43(3): 684–692
[7] VINYALS O, BLUNDELL C, LILLICRAO T P, et al. Mathcing networks for one shot learning[C]//Neural Information Processing Systems. Barcelona, 2016: 3630–3638.
[8] LI W, XU J, HUO J, et al. Distribution consistency based covariance metric networks for few-shot learning[C]//Proceedings of the AAAI Conference on Artificial Intelligence. Honolulu, 2019: 8642–8649.
[9] 丁辰伟, 孙闽红, 官友廉. 基于小样本条件的雷达辐射源个体识别算法[J]. 弹箭与制导学报, 2021, 41(3): 25–30
[10] 陈宏. 小波变换下舰船噪声信号识别方法[J]. 舰船科学技术, 2021, 43(14): 10–42
CHEN H. Research on recognition method of ship noise signal based on wavelet transform[J]. Ship Science and Technology, 2021, 43(14): 10–42
[11] LI R, WU Z, JIA J, et al. Towards discriminative representation learning for speech emotion recognition[C]//Proceedings of the 28th International Joint Conference on Artificial Intelligence. AAAI Press. AAAI Press, 2019: 5060–5066.
[12] SNELL J, SWERSKY K, ZEMEL R. Prototypical networks for few-shot learning[J]. Advances in neural information processing systems, 2017: 4077–4087.
[13] SUNG F, YANG Y, ZHANG L, et al. Learning to compare: Relation network for few-shot learning[C]//IEEE conference on computer vision and pattern recognition. Salt Lake City, 2018: 1199–1208.
[14] SANTORO A, BARTUNOV S, BOTVINICK M, et al. Meta-learning with memory-augmented neural networks[C]//International conference on machine learning. New York City, 2016: 1842–1850.
[15] HOU R, CHANG H, MA B, et al. Cross attention network for few-shot classification[C]//The 33rd International Conference on Neural Information Processing Systems. Vancouver, 2019: 4003–4014.
[16] LI Y, TIAN X, GONG M, LIUY, et al. Deep domain generalization via conditional invariant adversarial networks[C]// European Conference on Computer Vision (ECCV). Germany, 2018: 624–639.
