UWBR Ground Target Recognition Method Based on Range Doppler Map and Adaptive Feature Selection Network

doi:10.12382/bgxb.2024.1057

Abstract

Abstract:

The impulse radio ultra-wideband radar (IR-UWBR) has insufficient target recognition capability under the conditions of small sample size and complex detection scenes.Regarding the above-mentioned issue,this paper proposes a moving target recognition method based on range-Doppler map and adaptive feature selection network (RDM-AFSN).An IR-UWBR Doppler information extraction model is established based on the analysis of the law of IR-UWBR receiving the echo signals in the slow time dimension.At the same time,the characteristics of the moving target range-Doppler map,which has large differences in image spatial features due to complex background information and many target types,are deeply analyzed,and a RDM-AFSN target recognition model based on coordinate soft threshold denoising module and spatial adaptive down-sampling layer is constructed.Experimental results demonstrate that the proposed model effectively improves the classification capability of moving targets under small sample sizes and achieves good recognition results for similar targets in different scenarios.Compared to the convolutional-recurrent deep network and image coding deep network commonly used for ground target recognition,the proposed RDM-AFSN improves recognition accuracy by 3.64% and 7.53%,respectively.

Key words: pulse ultra-wideband radar, range-Doppler map, adaptive feature selection network, ground target recognition

CLC Number:

TJ434.1

HUANG Wenyu, XIONG Gang, LI Longlong, ZHANG Shuning, YU Wenxian. UWBR Ground Target Recognition Method Based on Range Doppler Map and Adaptive Feature Selection Network[J]. Acta Armamentarii, 2025, 46(9): 241057-.

Figures/Tables 20

References 24

[1]	LI B C, YANG J Y, YANG Y, et al. Sign language/gesture recognition based on cumulative distribution density features using UWB radar[J]. IEEE Transactions on Instrumentation and Measurement, 2021, 70:2511113.
[2]	ROHMAN B P A, ANDRA M B, NISHIMOTO M. Through-the-wall human respiration detection using UWB impulse radar on hovering drone[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2021, 14:6572-6584.
[3]	VEDDER B, SVENSSON B J, VINTER J, et al. Automated testing of ultrawideband positioning for autonomous driving[J]. Journal of Robotics, 2020, 2020(1):9345360.
[4]	SHI C, ZHENG Z J, PAN J, et al. Multiple stationary human targets detection in through-wall UWB radar based on convolutional neural network[J]. Applied Sciences, 2022, 12(9):4720.
[5]	HOU L S, ZHANG S N, WANG C X, et al. Jamming recognition of carrier-free UWB cognitive radar based on MANet[J]. IEEE Transactions on Instrumentation and Measurement, 2023.
[6]	LI J C, ZHANG S N, ZHU L Z, et al. Carrier-free ultra-wideband sensor target recognition in the jungle environment[J]. Remote Sensing, 2024, 16(9):1549.
[7]	周文, 郝新红, 董二娃, 等. 基于滑动多周期FFT的调频引信抗扫频干扰方法[J]. 兵工学报, 2023, 44(6):1744-1753. doi: 10.12382/bgxb.2022.0119
	ZHOU W, HAO X H, DONG E W, et al. Anti-frequency sweeping jamming method for FM fuze using sliding multi-cycle FFT[J]. Acta Armamentarii, 2023, 44(6):1744-1753. (in Chinese) doi: 10.12382/bgxb.2022.0119
[8]	杨瑾, 郝新红, 周文, 等. 基于异值检测及干扰样本归零的调频引信抗扫频干扰方法[J]. 探测与控制学报, 2023, 45(5):22-26.
	YANG J, HAO X H, ZHOU W, et al. FM fuze anti sweep interference based on heterogeneity detection and zero interference samples[J]. Journal of Detection and Control, 2023, 45(5):22-26. (in Chinese)
[9]	刘冰, 郝新红, 周文, 等. 基于BAS-BPNN的调频无线电引信目标与扫频干扰识别方法[J]. 兵工学报, 2023, 44(8):2391-2403. doi: 10.12382/bgxb.2022.0248
	LIU B, HAO X H, ZHOU W, et al. Recognition method of target and sweep jamming signal for FM radio fuze based on BAS-BPNN[J]. Acta Armamentarii, 2023, 44(8):2391-2403. (in Chinese) doi: 10.12382/bgxb.2022.0248
[10]	周文, 郝新红, 杨瑾, 等. 基于多尺度色散熵与支持向量机的调频引信扫频干扰信号识别方法[J]. 探测与控制学报, 2023, 45(6):21-26.
	ZHOU W, HAO X H, YANG J, et al. A FM fuze frequency sweeping jamming signal identification method based on multi-scale dispersion entropy and SVM[J]. Journal of Detection and Control, 2023, 45(6):21-26. (in Chinese)
[11]	刘冰, 郝新红, 段俊法, 等. 基于经验模态分解特征的无线电引信抗扫频式干扰方法[J]. 北京理工大学学报, 2023, 43(12):1290-1297.
	LIU B, HAO X H, DUAN J F, et al. Anti frequency sweeping jamming method for radio fuze based on empirical mode decomposition (EMD) features[J]. Transactions of Beijing institute of Technology, 2023, 43(12):1290-1297. (in Chinese)
[12]	房安琪, 李蓉. 基于数据增强的侵彻引信准确层识别神经网络方法[J]. 探测与控制学报, 2022, 44(1):1-6.
	FANG A Q, LI R. Penetration fuze accurate layer recognition neural network data enhancement method[J]. Journal of Detection and Control, 2022, 44(1):1-6. (in Chinese)
[13]	李晓雄, 张淑宁, 赵惠昌, 等. 基于1 DC-CGAN和小波能量特征的引信小样本地形目标识别[J]. 兵工学报, 2022, 43(10):2545-2553.
	LI X X, ZHANG S N, ZHAO H C, et al. Identification of fuzzy small-sample terrain targets based on 1DC-CGAN and wavelet energy features[J]. Acta Armamentarii, 2022, 43(10):2545-2553. (in Chinese) doi: 10.12382/bgxb.2021.0505
[14]	BO D Y, SHI C, WANG L L, et al. Specformer:spectral graph neural networks meet transformers:arXiv:2303.01028[J/OL]. Ithaca,NY, US:Cornell University, 2023(2023-03-02). https://arxiv.org/abs/2303.01028.
[15]	ZHEN T, XIONG G, YU W X. Fractal filter networks for sea state classification using radar sea clutter data[C]// Proceedings of the 2024 3rd Asia Conference on Algorithms,Computing and Machine Learning.Shanghai, China: Association for Computing Machinery, 2024:323-327.
[16]	ZHU Y Y, ZHANG S N, LI X X, et al. Ground target recognition using carrier-free UWB radar sensor with a semi-supervised stacked convolutional denoising autoencoder[J]. IEEE Sensors Journal, 2021, 21(18):20685-20693.
[17]	XIONG G, CHEN D, XIONG Z Q. OIF-CBDN:orientation information fusion CBD network for UWB-IR vehicle classification[J]. IEEE Geoscience and Remote Sensing Letters, 2024, 21:3502805.
[18]	CHEN D, XIONG G, WANG L Z, et al. Variable length sequential iterable convolutional recurrent network for UWB-IR vehicle target recognition[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61:5102311.
[19]	李龙龙. 变化场景下基于多特征融合的超宽带近程雷达目标识别[D]. 上海: 上海交通大学, 2023.
	LI L L. Ultra-wideband short-range radar target recognition based on multi-feature fusion in changing scenes[D]. Shanghai: Shanghai Jiao Tong University, 2023. (in Chinese)
[20]	LI X X, ZHANG S N, ZHU Y Y, et al. Supervised contrastive learning for vehicle classification based on the IR-UWB radar[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60:5117312.
[21]	ZHU L Z, SUN Y Y, ZHANG S N. Multi-angle recognition of vehicles based on carrier-free UWB sensor and deep residual shrinkage learning[J]. IEEE Microwave and Wireless Components Letters, 2022, 32(7):927-930.
[22]	XU Z H, BAKER C J, POONI S. Range and Doppler cell migration in wideband automotive radar[J]. IEEE Transactions on Vehicular Technology, 2019, 68(6):5527-5536.
[23]	LIU Z, MAO H Z, WU C Y, et al. A convnet for the 2020s[C]//Proceedings of the 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition.New Orleans,LA, US:IEEE, 2022:11976-11986.
[24]	ZHAO M H, ZHONG S S, FU X Y, et al. Deep residual shrinkage networks for fault diagnosis[J]. IEEE Transactions on Industrial Informatics, 2019, 16(7):4681-4690.

参数	数值
信号采样率fs/GHz	16
发射信号中心频率f₀/GHz	3.9
发射信号脉宽/ns	0.65
天线主瓣宽度/(°)	90
脉冲重复周期/ns	75
相干积累指数	64
接收脉冲间隔T_r/ms	1.25
测速范围/(m·s^-1)	-15~15
测距范围/m	2.5~15.0
分辨率Δr·Δv/(m²·s^-1)	≈0.28

参数	数值
信号采样率fs/GHz	16
发射信号中心频率f₀/GHz	3.9
发射信号脉宽/ns	0.65
天线主瓣宽度/(°)	90
脉冲重复周期/ns	75
相干积累指数	64
接收脉冲间隔T_r/ms	1.25
测速范围/(m·s^-1)	-15~15
测距范围/m	2.5~15.0
分辨率Δr·Δv/(m²·s^-1)	≈0.28

模型	准确率/%	精确率/%	召回率/%	F1Score	参数量/10³
CBDN^[17]	94.22	94.35	94.22	0.9424	52
ICDN^[19]	90.33	90.52	90.33	0.9035	72
ResNet	90.42	90.41	90.42	0.9143	52
VGG	89.23	89.11	89.23	0.8924	48
DRSNet^[21]	94.41	94.42	94.41	0.9410	84
ConvNeXt^[22]	93.33	93.42	93.33	0.9330	45
RDM-AFSN-S	97.86	97.89	97.86	0.9786	48
RDM-AFSN-L	98.50	98.51	98.50	0.9850	10300

模型	准确率/%	精确率/%	召回率/%	F1Score	参数量/10³
CBDN^[17]	94.22	94.35	94.22	0.9424	52
ICDN^[19]	90.33	90.52	90.33	0.9035	72
ResNet	90.42	90.41	90.42	0.9143	52
VGG	89.23	89.11	89.23	0.8924	48
DRSNet^[21]	94.41	94.42	94.41	0.9410	84
ConvNeXt^[22]	93.33	93.42	93.33	0.9330	45
RDM-AFSN-S	97.86	97.89	97.86	0.9786	48
RDM-AFSN-L	98.50	98.51	98.50	0.9850	10300

模型	准确率/%	精确率/%	召回率/%	F1Score
CBDN^[17]	92.68	93.64	89.01	0.9037
ICDN^[19]	91.90	90.05	91.56	0.8979
DRSNet^[21]	93.50	93.62	93.40	0.9340
ConvNeXt^[22]	93.53	92.12	93.54	0.9218
RDM-AFSN-S	95.66	94.99	95.42	0.9516