基于深度生成對抗網(wǎng)絡的海雜波數(shù)據(jù)增強方法

丁斌; 夏雪; 梁雪峰

doi:10.11999/JEIT200447

基于深度生成對抗網(wǎng)絡的海雜波數(shù)據(jù)增強方法

doi: 10.11999/JEIT200447 cstr: 32379.14.JEIT200447

丁斌^1, ,,
夏雪²,
梁雪峰^{1, 3}

1.
西安文理學院西安 710065
2.
西安石油大學西安 710065
3.
西安電子科技大學西安 710071

基金項目: 西安市科技計劃(2019KJWL30)

詳細信息

作者簡介:
丁斌：男，1980年生，博士，高級工程師，研究方向為智能信息處理、圖像解譯與智慧遙感

夏雪：女，1985年生，博士，研究方向為信號與信息處理

梁雪峰：男，1973年生，博士，教授，博士生導師，研究方向為視覺認知計算(心理學)、計算機視覺、視覺大數(shù)據(jù)挖掘、智能算法

通訊作者:
丁斌　xadb2005@163.com

中圖分類號: TN959.72; TP391
計量
- 文章訪問數(shù): 2389
- HTML全文瀏覽量: 1449
- PDF下載量: 300
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2020-06-02
- 修回日期: 2021-02-27
- 網(wǎng)絡出版日期: 2021-03-04
- 刊出日期: 2021-07-10

Sea Clutter Data Augmentation Method Based on Deep Generative Adversarial Network

Bin DING^{1
, ,},
Xue XIA²,
Xuefeng LIANG^{1, 3}

1.
Xi’an University, Xi’an 710065, China
2.
Xi’an Shiyou University, Xi’an 710065, China
3.
Xidian University, Xi’an 710071, China

Funds: Xi’an Science and Technology Plan (2019KJWL30)

摘要

摘要: 海雜波數(shù)據(jù)稀缺，獲取海雜波數(shù)據(jù)成本高、周期長，極大地限制了海雜波特性研究及海洋遙感應用。該文主要研究了基于深度生成性對抗網(wǎng)絡(GAN)的海雜波數(shù)據(jù)生成方法，通過擴展傳統(tǒng)的GAN框架，形成了1維海雜波數(shù)據(jù)生成和鑒別模型，基于實測海雜波數(shù)據(jù)集，進行對抗網(wǎng)絡生成和鑒別模型訓練，分析了生成模型所生成的海雜波數(shù)據(jù)的幅度分布特性和時間、空間相關性?；趯崪y數(shù)據(jù)驗證了該方法能夠生成更多、更多樣、與真實海雜波數(shù)據(jù)分布相近的海雜波數(shù)據(jù)。
- 生成性對抗網(wǎng)絡 /
- 海雜波 /
- 幅度分布特性 /
- 時間相關性
Abstract: Due to the scarcity of sea clutter data, the high cost and long period of obtaining sea clutter data greatly limit the research of sea clutter characteristics and the application of ocean remote sensing. The method of sea clutter data generation based on the Generative Adversarial Networks (GAN) is studied. By extending the traditional GAN framework, a one-dimensional sea clutter data generation and identification model is formed. Based on the radar measured sea clutter data set, the generation and identification model training in the adversarial network is carried out. The amplitude distribution characteristics and time and spatial correlation of the sea clutter data generated by the model are analyzed. Based on the measured data, it is verified that the method can generate more sea clutter data with more variety, and similar distribution to the real sea clutter data.
- Generative Adversarial Networks(GAN) /
- Sea clutter /
- Amplitude distribution characteristics /
- Temporal correlation

HTML全文

圖 1 GAN結構示意圖

下載: 全尺寸圖片幻燈片

圖 2 海雜波數(shù)據(jù)對抗生成網(wǎng)絡結構

下載: 全尺寸圖片幻燈片

圖 3 單載頻發(fā)射信號，海雜波時域波形

下載: 全尺寸圖片幻燈片

圖 4 實測海雜波數(shù)據(jù)實部&生成海雜波數(shù)據(jù)實部

下載: 全尺寸圖片幻燈片

圖 5 實測海雜波數(shù)據(jù)虛部&生成海雜波數(shù)據(jù)虛部

下載: 全尺寸圖片幻燈片

圖 6 第100采樣點10000幀海雜波幅度

下載: 全尺寸圖片幻燈片

圖 7 生成海雜波數(shù)據(jù)(10000幀)雜波幅度圖

下載: 全尺寸圖片幻燈片

圖 8 第100采樣點10000幀海雜波幅度直方圖

下載: 全尺寸圖片幻燈片

圖 9 生成海雜波數(shù)據(jù)(10000幀)幅度直方圖

下載: 全尺寸圖片幻燈片

圖 10 直接生成回波幅度(模值)(10000幀)直方圖

下載: 全尺寸圖片幻燈片

圖 11 幅度分布擬合曲線

下載: 全尺寸圖片幻燈片

圖 12 實測海雜波數(shù)據(jù)&生成海雜波數(shù)據(jù)功率譜密度

下載: 全尺寸圖片幻燈片

圖 13 實測&生成海雜波數(shù)據(jù)時間相關系數(shù)

下載: 全尺寸圖片幻燈片

圖 14 實測海雜波&生成海雜波距離向第100采樣點處距離維空間相關系數(shù)

下載: 全尺寸圖片幻燈片

表 1 生成器、辨別器網(wǎng)絡參數(shù)

生成器網(wǎng)絡			判別器網(wǎng)絡
Layer	Act./Norm	Output shape	Layer	Act./Norm	Output shape
Fully Linear	ReLU		Conv1d	Leaky ReLU	64×2048
	BatchNorm1d	1×256	Conv1d	Leaky ReLU	128×512
Conv1d	ReLU	512×512	Conv1d	Leaky ReLU	256×128
Conv1d	ReLU	256×1024	Conv1d	Leaky ReLU	512×128
Conv1d	ReLU	128×1024	Conv1d	Leaky ReLU	1024×16
Conv1d	ReLU	64×4096	Fully Linear	sigmoid	1×1
Conv1d	Tanh	1×8192

下載: 導出CSV

參考文獻(17)

[1]	劉寧波, 董云龍, 王國慶, 等. X波段雷達對海探測試驗與數(shù)據(jù)獲取[J]. 雷達學報, 2019, 8(5): 656–667. doi: 10.12000/JR19089 LIU Ningbo, DONG Yunlong, WANG Guoqing, et al. Sea-detecting X-band radar and data acquisition program[J]. Journal of Radars, 2019, 8(5): 656–667. doi: 10.12000/JR19089
[2]	DING Hao, GUAN Jian, LIU Ningbo, et al. Modeling of heavy tailed sea clutter based on the generalized central limit theory[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(11): 1591–1595. doi: 10.1109/LGRS.2016.2596322
[3]	TITI G W and MARSHALL D F. The ARPA/NAVY mountaintop program: Adaptive signal processing for airborne early warning radar[C]. 1996 IEEE International Conference on Acoustics, Speech, and Signal Processing Conference Proceedings, Atlanta, USA, 1996: 1165–1168.
[4]	DROSOPOULOS A. Description of the OHGR database[R]. Technical Note 94–14, 1994.
[5]	GRECO M, STINCO P, GINI F, et al. Impact of sea clutter nonstationarity on disturbance covariance matrix estimation and CFAR detector performance[J]. IEEE Transactions on Aerospace and Electronic Systems, 2010, 46(3): 1502–1513. doi: 10.1109/TAES.2010.5545205
[6]	王帥. 基于人工智能(GAN)的影像技術探究[D]. [碩士論文], 南京師范大學, 2019.
[7]	雷志勇, 黃忠平, 吳剛, 等. 機載L波段雷達海雜波幅度分布特性分析[J]. 電波科學學報, 2019, 34(5): 558–566. LEI Zhiyong, HUANG Zhongping, WU Gang, et al. Analysis of sea clutter distribution with L-band airborne radar[J]. Chinese Journal of Radio Science, 2019, 34(5): 558–566.
[8]	劉恒燕, 宋杰, 熊偉, 等. 大入射余角海雜波相關特性分析及幅度擬合[J]. 海軍航空工程學院學報, 2018, 33(3): 307–312. doi: 10.7682/j.issn.1673-1522.2018.03.009 LIU Hengyan, SONG Jie, XIONG Wei, et al. Sea clutter correlation analysis and amplitude fitting for large grazing angle[J]. Journal of Naval Aeronautical and Astronautical University, 2018, 33(3): 307–312. doi: 10.7682/j.issn.1673-1522.2018.03.009
[9]	傅俊滔, 周國安, 陳紅. 基于ZMNL的Pareto雜波模擬改進方法[J]. 彈箭與制導學報, 2019, 39(4): 19–21, 28. FU Juntao, ZHOU Guoan, and CHEN Hong. Improved method of Pareto clutter simulation based on ZMNL[J]. Journal of Projectiles,Rockets,Missiles and Guidance, 2019, 39(4): 19–21, 28.
[10]	王坤峰, 左旺孟, 譚營, 等. 生成式對抗網(wǎng)絡: 從生成數(shù)據(jù)到創(chuàng)造智能[J]. 自動化學報, 2018, 44(5): 769–774. WANG Kunfeng, ZUO Wangmeng, TAN Ying, et al. Generative adversarial networks: from generating data to creating intelligence[J]. Acta Automatica Sinica, 2018, 44(5): 769–774.
[11]	徐雅楠, 劉寧波, 丁昊, 等. 利用CNN的海上目標探測背景分類方法[J]. 電子學報, 2019, 47(12): 2505–2514. XU Yanan, LIU Ningbo, DING Hao, et al. Background classification method for marine target detection based on CNN[J]. Acta Electronica Sinica, 2019, 47(12): 2505–2514.
[12]	丁昊, 劉寧波, 董云龍, 等. 雷達海雜波測量試驗回顧與展望[J]. 雷達學報, 2019, 8(3): 281–302. doi: 10.12000/JR19006 DING Hao, LIU Ningbo, DONG Yunlong, et al. Overview and prospects of radar sea clutter measurement experiments[J]. Journal of Radars, 2019, 8(3): 281–302. doi: 10.12000/JR19006
[13]	GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial networks[C]. Conference on Neural Information Processing Systems (NIPS), Montreal Canada, 2014: 1–23.
[14]	ZHANG Zhimian, WANG Haipeng, XU Feng, et al. Complex-valued convolutional neural network and its application in polarimetric SAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(12): 7177–7188.
[15]	ARJOVSKY M, CHINTALA S, and BOTTOU L. Wasserstein GAN[C]. International Conference on Machine Learning, Sydneym, Australia, 2017: 1–32.
[16]	GULRAJANI I, AHMED F, ARJOVSKY M, et al. Improved training of Wasserstein GANs[C]. The 31st International Conference on Neural Information Processing Systems, Long Beach, USA, 2017: 5767–5777.
[17]	關鍵, 丁昊, 黃勇, 等. 實測海雜波數(shù)據(jù)空間相關性研究[J]. 電波科學學報, 2012, 27(5): 943–953. GUAN Jian, DING Hao, HUANG Yong, et al. Spatial correlation property with measured sea clutter data[J]. Chinese Journal of Radio Science, 2012, 27(5): 943–953.