基于迁移学习的雷达杂波幅度统计模型选择

doi:10.12305/j.issn.1001-506X.2022.08.09

摘要/Abstract

摘要：

在对目标进行探测和识别的过程中, 雷达杂波幅度统计模型选择是重要步骤。为了提升杂波幅度统计模型选择的准确率, 基于样本分布适配, 提出了一种加权再均衡分布适配的迁移学习方法, 实现了仿真数据的信息向实测海杂波IPIX数据的迁移。通过与已有算法进行比较, 实验结果表明改进后的算法在IPIX数据集上能取得更好的分类准确率, 在迁移学习公共数据集Office-Caltech10上的验证结果也表明了算法的普适性。

关键词: 雷达杂波, IPIX数据, 迁移学习, 加权再均衡分布适配

Abstract:

In the process of target detection and recognition, the selection of a radar clutter amplitude statistical model is an important step. In order to improve the accuracy of clutter amplitude statistical model selection, based on the adaptive sample distribution, a weighted rebalance distribution adaptation transfer learning method is proposed to realize the information transfer from simulation data to the measured sea clutter IPIX data. Compared with the existing algorithms, the experimental results show that the improved algorithm can achieve better classification accuracy, and the verification results on the transfer learning public dataset Office-Caltech10 also show the universality of the algorithm.

Key words: radar clutter, IPIX date, transfer learning, weighted rebalance distribution adaptation

中图分类号:

TN957.51

杨立儒, 刘永祥, 杨威. 基于迁移学习的雷达杂波幅度统计模型选择[J]. 系统工程与电子技术, 2022, 44(8): 2457-2467.

Liru YANG, Yongxiang LIU, Wei YANG. Radar clutter amplitude statistical model selection based on transfer learning[J]. Systems Engineering and Electronics, 2022, 44(8): 2457-2467.

图/表 11

图1

表1

图2

表2

图3

图4

表3

表4

表5

图5

图6

参考文献 36

1	席现国. 雷达杂波的恒虚警处理的研究[D]. 大连: 大连海事大学, 2013.
	XI X G. Study on CFAR in radar clutter[D]. Dalian: Dalian Maritime University, 2013.
2	宋希珍. 海杂波幅度分布参数矩估计和恒虚警检测[D]. 西安: 西安电子科技大学, 2014.
	SONG X Z. Moment estimation of amplitude distribution parameters of sea clutter and CFAR detection[D]. Xi'an: Xidian University, 2014.
3	方学立, 马晓岩, 向家彬. 基于高阶统计量特征提取的雷达杂波分类识别[J]. 空军雷达学院学报, 2000, 14 (1): 9- 12.
	FANG X L , MA X Y , XIANG J B . Radar clutter recognition and higher-order statistics feature extraction[J]. Journal of the Air Force Radar Academy, 2000, 14 (1): 9- 12.
4	饶辉. 低分辨雷达海上目标分类与杂波特性研究[D]. 西安: 西安电子科技大学, 2015.
	RAO H. Study of sea targets classification and clutter characteristics based on low-resolution radar[D]. Xi'an: Xidian University, 2011.
5	M.W.朗, 薛德镛. 陆地和海面的雷达波散射特性[M]. 北京: 科学出版社, 1981: 173- 177.
	LANG M W , XUE D Y . Radar reflectivity of land and sea[M]. Beijing: Science Press, 1981: 173- 177.
6	R.L.米切尔, 陈训达. 雷达系统模拟[M]. 北京: 科学出版社, 1982: 49- 50.
	MITCHELL R L , CHEN X D . Radar system simulation[M]. Beijing: Science Press, 1982: 49- 50.
7	申玉, 陶然, 单涛. 相关对数正态分布雷达杂波的建模与仿真[J]. 火控雷达技术, 2001, 30 (4): 1- 5. doi: 10.3969/j.issn.1008-8652.2001.04.001
	SHEN Y , TAO R , SHAN T . Modeling and simulation of correlated log-normal distribution radar clutter[J]. Fire control radar technology, 2001, 30 (4): 1- 5. doi: 10.3969/j.issn.1008-8652.2001.04.001
8	CHAN H C . Radar sea-clutter at low grazing angles[J]. Radar & Signal Processing IEE Proceedings F, 1990, 137 (2): 102- 112.
9	WALKER D . Doppler modelling of radar sea clutter[J]. IEE Proceedings-Radar, Sonar and Navigation, 2002, 148 (2): 73- 80.
10	WATTS S . Radar Detection Prediction in K-Distributed Sea Clutter and Thermal Noise[J]. IEEE Trans.on Aerospace and Electronic Systems, 1987, AES-23 (1): 40- 45. doi: 10.1109/TAES.1987.313334
11	杨海文. 海杂波建模与实测数据分析[D]. 西安: 西安电子科技大学, 2011.
	YANG H W. Modeling of sea clutter and analysis of measured data[D]. Xi'an: Xidian University, 2011.
12	XIAO Y H , LIU W J , GAO L P . Radar signal recognition based on transfer learning and feature fusion[J]. Mobile Networks and Applications, 2020, 25, 1563- 1571. doi: 10.1007/s11036-019-01360-1
13	GENG J , DENG X Y , MA X R , et al. Transfer learning for SAR image classification via deep joint distribution adaptation networks[J]. IEEE Trans.on Geoscience and Remote Sensing, 2020, 58 (8): 5377- 5392. doi: 10.1109/TGRS.2020.2964679
14	HUANG Z L , DUMITRU C O , PAN Z X , et al. Classification of large-scale high-resolution SAR images with deep transfer learning[J]. IEEE Geoscience and Remote Sensing Letters, 2020, PP (99): 1- 5.
15	MA Y , CAMPBELL A T , COOK D J , et al. Transfer learning for activity recognition in mobile health[J]. DSHealth, 2020,
16	BROPHY E, MUEHLHAUSEN W, SMEATON A F, et al. Optimised convolutional neural networks for heart rate estimation and human activity recognition in wrist worn sensing applications[C]//Proc. of the IEEE Conference on Pervasive Computing and Communications Workshops, 2020.
17	FALCO P , LU S , NATALE C , et al. A transfer learning approach to cross-modal object recognition: from visual observation to robotic haptic exploration[J]. IEEE Trans.on Robotics, 2020, 35 (4): 987- 999.
18	MONKA S , HALILAJ L , SCHMID S , et al. ConTraKG: contrastive-based transfer learning for visual object recognition using knowledge graphs[J]. Computer Vision and Pattern Recognition, 2021,
19	RIVERA S , MENDOZA-SCHROCK O , DIEHL A . Transfer learning for aided target recognition: comparing deep learning to other machine learning approaches[J]. Computer Vision and Pattern Recognition, 2020,
20	杨强. 迁移学习[M]. 北京: 机械工业出版社, 2020.
	YANG Q . Transfer learning[M]. Beijing: China Machine Press, 2020.
21	GRETTON A, SCHÖLKOPF B, HUANG J. Correcting sample selection bias by unlabeled data[C]//Proc. of the Conference on Advances in Neural Information Processing Systems, 2007: 601-608.
22	JAIN S , SIMON H U , TOMITA E . Measuring statistical dependence with Hilbert-Schmidt norms[J]. Computer Science, 2005, 3734, 63- 77.
23	PAN S J, KWOK J T, YANG Q. Transfer learning via dimensionality reduction[C]//Proc. of the 23rd AAAI Conference on Artificial Intelligence, 2008: 13-17.
24	PAN S J , TSANG I W , KWOK J T , et al. Domain adaptation via transfer component analysis[J]. IEEE Trans.on Neural Networks, 2011, 22 (2): 199- 210. doi: 10.1109/TNN.2010.2091281
25	SATPAL S, SARAWAGI S. Domain adaptation of conditional probability models via feature subsetting[C]//Proc. of the European Conference on Principles & Practice of Knowledge Discovery in Databases, 2007: 224-235.
26	LONG M S, WANG J M, DING G G, et al. Transfer feature learning with joint distribution adaptation[C]//Proc. of the International Conference on Computer Vision, 2013: 2200-2207.
27	HOU C A , TSAI Y H H , YEH Y R , et al. Unsupervised domain adaptation with label and structural consistency[J]. IEEE Trans.on Image Processing, 2016, 5552- 5562.
28	WANG J D, CHEN Y Q, HAO S J, et al. Balanced distribution adaptation for transfer learning[C]//Proc. of the International Conference on Date Mining, 2017: 1129-1134.
29	任博. 基于实测数据的地海杂波特性研究及仿真应用[D]. 长沙: 国防科学技术大学, 2011.
	REN B. Research and simulation application of ground sea clutter characteristics based on measured data[D]. Changsha: National University of Defense Technology, 2011.
30	邓怀泽. 基于实测数据的海杂波统计建模[D]. 西安: 西安电子科技大学, 2014.
	DENG H Z. Statistical modeling of sea clutter based on measured data[D]. Xi'an: Xidian University, 2014.
31	MENON M V . Estimation of the shape and scale parameters of the Weibull distribution[J]. Technometrics, 1963, 15, 175- 182.
32	石志广. 基于统计与复杂性理论的杂波特性分析及信号处理方法研究[D]. 长沙: 国防科学技术大学, 2007.
	SHI Z G. Clutter analysis and signal processing based on statistics and complexity theory[D]. Changsha: National University of Defense Technology, 2007.
33	李芾. 基于实测海杂波数据的统计特性分析[D]. 西安: 西安电子科技大学, 2013.
	LI F. Statistical characteristics analysis based on measured sea clutter data[D]. Xi'an: Xidian University, 2013.
34	宋希珍. 海杂波幅度分布参数矩估计和恒虚警检测[D]. 西安: 西安电子科技大学, 2014.
	SONG X Z. Moment estimation of amplitude distribution parameters of sea clutter and CFAR detection[D]. Xi'an: Xidian University, 2014.
35	时公涛, 赵凌君, 桂琳, 等. 基于Mellin变换的K分布参数估计新方法[J]. 电子学报, 2010, 38 (9): 2083- 2089.
	SHI G T , ZHAO L J , GUI L . A novel parameter estimation method for the K distribution based on the Mellin transform[J]. Acta Electronica Sinica, 2010, 38 (9): 2083- 2089.
36	时公涛. 基于干涉图的多通道SAR地面慢动目标自动检测技术研究[D]. 长沙: 国防科学技术大学, 2009.
	SHI G T. The research on multi-channel SAR automatic detection of slow ground moving targets based on the interferogram[D]. Changsha: National University of Defense Technology, 2009.

类型	对数正态		韦布尔		K		瑞利
类型	σ	μ	q	p	v	a	σ
标签	1		2		3		4
图 1(a)	0.553	-0.505	0.984	7.523	1.317	1.617	1.003
图 1(b)	0.569	-0.718	0.999	2.466	1.946	1.483	0.812

任务类型	迁移任务	1NN	TCA+1NN	W-BDA+1NN	W-RBDA+1NN
均匀样本	Xs80→Xt360(小样本)	48.89	71.94	82.22	88.89
	Xs200→ Xt360(小样本)	49.72	72.22	91.39	93.61
	Xs400 →Xt360(等量样本)	53.89	71.11	89.44	90.56
	Xs1200 → Xt360(大样本)	51.67	72.22	90.28	91.11
非均匀样本	Xs80 → Xt200	33.50	56.00	84.50	87.00
	Xs200 →Xt200	34.00	56.00	86.00	91.00
	Xs400 →Xt200	37.00	57.50	80.00	88.00
	Xs1200→Xt200	33.50	58.50	84.50	92.50

样本	对数正态	韦布尔	K	瑞利	标签
1	0.004 4	0.000 8	0.000 4	0.003 0	3
2	0.008 0	0.036 6	0.023 6	0.066 8	1
3	0.013 2	0.002 4	0.188 9	0.017 1	2
4	0.209 2	0.023 9	2.727 8	0.006 7	4

样本	标签	对数正态	韦布尔	K	瑞利
1NN	1	0.044 4	0.016 2	0.424 0	0.067 6
	2	0.183 7	0.012 2	2.092 0	0.052 0
	3	0.003 0	0.000 4	0.000 2	0.002 7
	4	0.047 9	0.006 0	0.602 8	0.146 6
TCA	1	0.007 2	0.022 7	0.028 6	0.107 5
	2	0.195 8	0.012 9	2.257 7	0.059 7
	3	0.052 5	0.005 3	0.480 4	0.014 1
	4	0.026 2	0.006 1	0.183 3	0.045 7
W-BDA	1	0.060 0	0.022 7	0.644 8	0.096 5
	2	0.084 3	0.004 8	1.015 3	0.107 0
	3	0.025 9	0.005 2	0.167 6	0.018 2
	4	0.281 0	0.019 8	3.185 2	0.013 2
W-RBDA	1	0.012 7	0.022 5	0.081 4	0.106 2
	2	0.115 0	0.006 3	1.375 1	0.101 0
	3	0.048 8	0.004 1	0.440 8	0.011 6
	4	0.269 8	0.020 1	3.047 6	0.011 1

迁移任务	W-BDA+1NN	W-RBDA+1NN	迁移任务	W-BDA+1NN	W-RBDA+1NN
C→A	45.30	48.23	W→C	35.00	35.80
C→W	39.66	40.00	W→A	38.62	39.04
C→D	50.96	50.96	W→D	92.36	92.36
A→C	40.69	40.69	D→C	33.75	34.19
A→W	41.69	41.69	D→A	36.53	36.95
A→D	39.49	40.13	D→W	90.51	90.51