为提高雷达高分辨距离像船只目标识别准确率，提出一种基于双分支特征融合卷积神经网络的船只目标识别方法。设计了提取不同层次特征的2个分支。使用堆叠卷积层结构且降采样次数少的细节分支，提取船只的高分辨率局部特征；使用模块化结构组成全局分支，提取船只的低分辨率全局姿态特征。根据特征图通过2个分支后的维度变化，在特征融合模块中对2种特征进行尺寸调整，相互融合并输出识别结果。实验结果显示，所提方法与传统识别方法相比，收敛更快、参数量更少且具有更高的准确性，验证了其在高分辨距离像船只分类任务上的有效性。
To improve the accuracy of radar high resolution range profile ship target recognition, a ship target recognition method based on dual-branch feature fusion convolutional neural network model is proposed. Two branches are designed to extract features at different levels. The method designs a stacked convolutional detail branch with reduced downsampling to extract high resolution local features of ships. The global branch is composed of a modular structure used to extract low resolution global attitude features of ships. Based on the dimensional changes of the feature map after passing through two branches, the two features are changed in size separately in the feature fusion module, and the features are fused with each other to output recognition results. The experimental results show that the proposed method has faster convergence, fewer parameters, and higher accuracy compared to traditional recognition methods, verifying its effectiveness in HRRP ship classification.
[1] Li L, Liu Z, Li T. Radar high-resolution range profile feature extraction method based on multiple kernel projection subspace fusion[J]. IET Radar, Sonar & Navigation, 2018, 12(4): 417-425. DOI: 10.1049/iet-rsn.2017.0391.
[2] 韩磊, 姚璐. HRRP自动目标识别方法综述[J]. 北京理工大学学报, 2020, 40(4): 351-361, 381. DOI: 10.15918/j.tbit1001-0645.2019.057.
[3] Bouarara H A. Recurrent neural network (RNN) to analyse mental behaviour in social media[J]. International Journal of Software Science and Computational Intelligence, 2021, 13(3): 1-11. DOI: 10.4018/ijssci.2021070101.
[4] Gruber N. Detecting dynamics of action in text with a recurrent neural network[J]. Neural Computing and Applications, 2021, 33(22): 15709-15718. DOI: 10.1007/s00521-021-06190-5.
[5] Li J Q, Li S J, Liu Q, et al. A novel algorithm for HRRP target recognition based on CNN[C]//International Conference on Internet of Things as a Service. Cham: Springer, 2020: 397-404.10.1007/978-3-030-44751-9_33.
[6] 王容川, 庄志洪, 王宏波, 等. 基于卷积神经网络的雷达目标HRRP分类识别方法[J]. 现代雷达, 2019, 41(5): 33-38. DOI: 10.16592/j.cnki.1004-7859.2019.05.007.
[7] Lin C L, Chen T P, Fan K C, et al. Radar high-resolution range profile ship recognition using two-channel convolutional neural networks concatenated with bidirectional long short-term memory[J]. Remote Sensing, 2021, 13(7): 1259. DOI: 10.3390/rs13071259.
[8] Lundén J, Koivunen V. Deep learning for HRRP-based target recognition in multistatic radar systems[C]//2016 IEEE Radar Conference (RadarConf). Philadelphia, PA, USA. IEEE, 2016: 1-6. DOI: 10.1109/RADAR.2016.7485271.
[9] Diao Y J, Liu S W, Gao X Z, et al. CNN based on multiscale window self-attention mechanism for radar HRRP target recognition[C]//2022 7th International Conference on Signal and Image Processing (ICSIP). Suzhou, China. IEEE, 2022: 281-285. DOI: 10.1109/ICSIP55141.2022.9887058.
[10] 郭晨, 简涛, 徐从安, 等. 基于深度多尺度一维卷积神经网络的雷达舰船目标识别[J]. 电子与信息学报, 2019, 41(6): 1302-1309. DOI: 10.11999/JEIT180677.
[11] Karabayır O, Yücedağ O M, Kartal M Z, et al. Convolutional neural networks-based ship target recognition using high resolution range profiles[C]//2017 18th International Radar Symposium (IRS). Prague, Czech Republic. IEEE, 2017: 1-9. DOI: 10.23919/IRS.2017. 8008207.
[12] Liu J Q, Chen B, Chen W C, et al. Radar HRRP target recognition with target aware two-dimensional recurrent neural network[C]//2019 IEEE International Conference on Signal Processing, Communications and Computing (ICSPCC). Dalian, China. IEEE, 2019: 1-6. DOI: 10.1109/ICSPCC46631.2019.8960736.
[13] Pan M, Liu A L, Yu Y Z, et al. Radar HRRP target recognition model based on a stacked CNN-Bi-RNN with attention mechanism[J]. IEEE Transactions on Geoscience and Remote Sensing, 1809, 60: 5100814. DOI: 10.1109/TGRS.2021.3055061.
[14] 郭帅, 陈婷, 王鹏辉, 等. 基于角度引导Transformer融合网络的多站协同目标识别方法[J]. 雷达学报, 2023, 12(3): 516-528. DOI: 10.12000/JR23014.
[15] Xu E C, Xiong G, Zhang S N, et al. HRRP target recognition based on the dual-mode gram angle field features and the multi-level CNN[C]//2021 IEEE 6th International Conference on Signal and Image Processing (ICSIP). Nanjing, China. IEEE, 2021: 77-81. DOI: 10.1109/ICSIP52628.2021.9688620.
[16] 冯国徽. 基于卷积神经网络VGG模型的小规模图像分类[D]. 兰州: 兰州大学, 2018.
[17] Chollet F. Xception: Deep learning with depthwise separable convolutions[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu, HI, USA. IEEE, 2017: 1800-1807. DOI: 10.1109/CVPR.2017.195.
[18] Howard A, Sandler M, Chen B, et al. Searching for MobileNetV3[C]//2019 IEEE/CVF International Conference on Computer Vision (ICCV). Seoul, Korea (South). IEEE, 2019: 1314-1324. DOI: 10.1109/ICCV.2019.00140.
[19] Zhang X Y, Zhou X Y, Lin M X, et al. ShuffleNet: An extremely efficient convolutional neural network for mobile devices[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, USA. IEEE, 2018: 6848-6856. DOI: 10.1109/CVPR.2018.00716.
[20] 周文雪, 张华春. 一种面向SAR图像快速舰船检测的轻量化网络[J]. 中国科学院大学学报, 2024,41(6)： 776-785. DOI: 10.7523/j.ucas.2023.017.
[21] Hou X Y, Ao W, Song Q, et al. FUSAR-ship: building a high-resolution SAR-AIS matchup dataset of Gaofen-3 for ship detection and recognition[J]. Science China Information Sciences, 2020, 63(4): 140303. DOI: 10.1007/s11432-019-2772-5.
[22] Liu S Q, Zhan R H, Zhai Q L, et al. Multi-view radar target recognition based on Bayesian compressive sensing[J]. Applied Mechanics and Materials, 2014, 687/688/689/690/691: 3704-3708. DOI: 10.4028/www.scientific.net/amm.687-691.3704.
[23] 刘巧圆, 郭宝峰, 王刚, 等. 针对舰船目标高分辨距离像的特征选择方法[J]. 火力与指挥控制, 2017, 42(8): 82-86. DOI: 10.3969/j.issn.1002-0640.2017.08.019.
[24] 李月琴, 张红莉, 张维, 等. 基于改进CNN的HRRP目标识别方法[J]. 兵器装备工程学报, 2022, 43(8): 265-274. DOI: 10.11809/bqzbgcxb2022.08.041.
[25] 岳智彬, 卢建斌, 万露. 基于注意力机制的SRU模型雷达HRRP目标识别[J]. 舰船电子工程, 2023, 43(4): 44-48. DOI: 10.3969/j.issn.1672-9730.2023.04.010.
