基于半監(jiān)督學(xué)習(xí)的SAR目標(biāo)檢測網(wǎng)絡(luò)

杜蘭; 魏迪; 李璐; 郭昱辰

doi:10.11999/JEIT190783

基于半監(jiān)督學(xué)習(xí)的SAR目標(biāo)檢測網(wǎng)絡(luò)

doi: 10.11999/JEIT190783 cstr: 32379.14.JEIT190783

西安電子科技大學(xué)雷達(dá)信號(hào)處理國家重點(diǎn)實(shí)驗(yàn)室西安 710071

基金項(xiàng)目: 國家自然科學(xué)基金(61771362, U1833203, 61671354)，高等學(xué)校學(xué)科創(chuàng)新引智計(jì)劃(B18039)，陜西省重點(diǎn)科技創(chuàng)新團(tuán)隊(duì)計(jì)劃

詳細(xì)信息

作者簡介:
杜蘭：女，1980年生，教授，博士生導(dǎo)師，研究方向?yàn)榻y(tǒng)計(jì)信號(hào)處理、雷達(dá)信號(hào)處理、機(jī)器學(xué)習(xí)及其在雷達(dá)目標(biāo)檢測與識(shí)別方面的應(yīng)用

魏迪：男，1995年生，碩士生，研究方向?yàn)槔走_(dá)目標(biāo)檢測與識(shí)別，機(jī)器學(xué)習(xí)等

李璐：女，1992年生，博士生，研究方向?yàn)闄C(jī)器學(xué)習(xí)和目標(biāo)檢測識(shí)別等

郭昱辰：男，1992年生，博士生，研究方向?yàn)槔走_(dá)目標(biāo)檢測與識(shí)別

通訊作者:
杜蘭　dulan@mail.xidian.edu.cn

中圖分類號(hào): TN957.51
計(jì)量
- 文章訪問數(shù): 5081
- HTML全文瀏覽量: 2385
- PDF下載量: 371
- 被引次數(shù): 0
出版歷程
- 收稿日期: 2019-10-12
- 修回日期: 2019-12-05
- 網(wǎng)絡(luò)出版日期: 2019-12-09
- 刊出日期: 2020-01-21

SAR Target Detection Network via Semi-supervised Learning

National Laboratory of Radar Signal Processing, Xidian University, Xi’an 710071, China

Funds: The National Science Foundation of China (61771362, U1833203, 61671354), 111 Project (B18039), Shaanxi Provience Innovation Team Project

摘要

摘要:
現(xiàn)有的基于卷積神經(jīng)網(wǎng)絡(luò)(CNN)的合成孔徑雷達(dá)(SAR)圖像目標(biāo)檢測算法依賴于大量切片級標(biāo)記的樣本，然而對SAR圖像進(jìn)行切片級標(biāo)記需要耗費(fèi)大量的人力和物力。相對于切片級標(biāo)記，僅標(biāo)記圖像中是否含有目標(biāo)的圖像級標(biāo)記較為容易。該文利用少量切片級標(biāo)記的樣本和大量圖像級標(biāo)記的樣本，提出一種基于卷積神經(jīng)網(wǎng)絡(luò)的半監(jiān)督SAR圖像目標(biāo)檢測方法。該方法的目標(biāo)檢測網(wǎng)絡(luò)由候選區(qū)域提取網(wǎng)絡(luò)和檢測網(wǎng)絡(luò)組成。半監(jiān)督訓(xùn)練過程中，首先使用切片級標(biāo)記的樣本訓(xùn)練目標(biāo)檢測網(wǎng)絡(luò)，訓(xùn)練收斂后輸出的候選切片構(gòu)成候選區(qū)域集；然后將圖像級標(biāo)記的雜波樣本輸入網(wǎng)絡(luò)，將輸出的負(fù)切片加入候選區(qū)域集；接著將圖像級標(biāo)記的目標(biāo)樣本也輸入網(wǎng)絡(luò)，對輸出結(jié)果中的正負(fù)切片進(jìn)行挑選并加入候選區(qū)域集；最后使用更新后的候選區(qū)域集訓(xùn)練檢測網(wǎng)絡(luò)。更新候選區(qū)域集和訓(xùn)練檢測網(wǎng)絡(luò)交替迭代直至收斂?；趯?shí)測數(shù)據(jù)的實(shí)驗(yàn)結(jié)果證明，所提方法的性能與使用全部樣本進(jìn)行切片級標(biāo)記的全監(jiān)督方法的性能相差不大。
- 合成孔徑雷達(dá) /
- 目標(biāo)檢測 /
- 半監(jiān)督學(xué)習(xí) /
- 卷積神經(jīng)網(wǎng)絡(luò)
Abstract:
The current Synthetic Aperture Radar (SAR) target detection methods based on Convolutional Neural Network (CNN) rely on a large amount of slice-level labeled train samples. However, it takes a lot of labor and material resources to label the SAR images at slice-level. Compared to label samples at slice-level, it is easier to label them at image-level. The image-level label indicates whether the image contains the target of interest or not. In this paper, a semi-supervised SAR image target detection method based on CNN is proposed by using a small number of slice-level labeled samples and a large number of image-level labeled samples. The target detection network of this method consists of region proposal network and detection network. Firstly, the target detection network is trained using the slice-level labeled samples. After training convergence, the output slices constitute the candidate region set. Then, the image-level labeled clutter samples are input into the network and then the negative slices of the output are added to the candidate region set. Next, the image-level labeled target samples are input into the network as well. After selecting the positive and negative slices in the output of the network, they are added to the candidate region set. Finally, the detection network is trained using the updated candidate region set. The processes of updating candidate region set and training detection network alternate until convergence. The experimental results based on the measured data demonstrate that the performance of the proposed method is similar to the fully supervised training method using a much larger set of slice-level samples.
- SAR /
- Target detection /
- Semi-supervised learning /
- Convolutional Neural Network (CNN)

HTML全文

圖 1 半監(jiān)督SAR圖像目標(biāo)檢測方法

下載: 全尺寸圖片幻燈片

圖 2 特征提取網(wǎng)絡(luò)

下載: 全尺寸圖片幻燈片

圖 3 數(shù)據(jù)集示例

下載: 全尺寸圖片幻燈片

圖 4 MiniSAR數(shù)據(jù)集：Gaussian-CFAR的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 5 MiniSAR數(shù)據(jù)集：Faster R-CNN-少部分切片級標(biāo)記的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 6 MiniSAR數(shù)據(jù)集：Faster R-CNN-全部切片級標(biāo)記的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 7 MiniSAR數(shù)據(jù)集：文獻(xiàn)[14]方法的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 8 MiniSAR數(shù)據(jù)集：文獻(xiàn)[15]方法的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 9 MiniSAR數(shù)據(jù)集：本文方法的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 10 FARADSAR數(shù)據(jù)集：Gaussian-CFAR的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 11 FARADSAR數(shù)據(jù)集：Faster R-CNN-少部分切片級標(biāo)記的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 12 FARADSAR數(shù)據(jù)集：Faster R-CNN-全部切片級標(biāo)記的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 13 FARADSAR數(shù)據(jù)集：文獻(xiàn)[14]方法的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 14 FARADSAR數(shù)據(jù)集：文獻(xiàn)[15]方法的檢測結(jié)果

下載: 全尺寸圖片幻燈片

圖 15 FARADSAR數(shù)據(jù)集：本文方法的檢測結(jié)果

下載: 全尺寸圖片幻燈片

表 1 不同方案的實(shí)驗(yàn)結(jié)果

負(fù)包數(shù)量	挑選的切片	$P$	$R$	F1-score
0	正切片	0.6397	0.7500	0.6905
	負(fù)切片	0.8833	0.4569	0.6023
	正切片+負(fù)切片	0.7387	0.7069	0.7225
10	正切片	0.6797	0.7500	0.7131
	負(fù)切片	0.7917	0.4914	0.6064
	正切片+負(fù)切片	0.7573	0.6724	0.7123
20	正切片	0.7658	0.7328	0.7489
	負(fù)切片	0.8382	0.4914	0.6196
	正切片+負(fù)切片	0.8137	0.7155	0.7615
30	正切片	0.8202	0.6293	0.7122
	負(fù)切片	0.8413	0.4569	0.5922
	正切片+負(fù)切片	0.8675	0.6207	0.7236
40	正切片	0.8111	0.6293	0.7087
	負(fù)切片	0.8667	0.4483	0.5909
	正切片+負(fù)切片	0.8352	0.6552	0.7343

下載: 導(dǎo)出CSV

表 2 不同方法的實(shí)驗(yàn)結(jié)果

不同方法	MiniSAR數(shù)據(jù)集			FARADSAR數(shù)據(jù)集
不同方法	$P$	$R$	F1-score	$P$	$R$	F1-score
Gaussian-CFAR	0.3789	0.7966	0.5135	0.2813	0.4671	0.3512
Faster R-CNN-少部分切片級標(biāo)記	0.6455	0.6121	0.6283	0.7370	0.8813	0.8027
Faster R-CNN-全部切片級標(biāo)記	0.8073	0.7586	0.7822	0.7760	0.9479	0.8534
文獻(xiàn)[14]方法	0.5814	0.9806	0.7285	0.4506	0.7325	0.5580
文獻(xiàn)[15]方法	0.4699	0.7480	0.5772	0.3744	0.7945	0.5090
本文方法	0.8137	0.7155	0.7615	0.8035	0.8813	0.8406

下載: 導(dǎo)出CSV

參考文獻(xiàn)(19)

NOVAK L M, BURL M C, and IRVING W W. Optimal polarimetric processing for enhanced target detection[J]. IEEE Transactions on Aerospace and Electronic Systems, 1993, 29(1): 234–244. doi: 10.1109/7.249129

XING X W, CHEN Z L, ZOU H X, et al. A fast algorithm based on two-stage CFAR for detecting ships in SAR images[C]. The 2nd Asian-Pacific Conference on Synthetic Aperture Radar, Xi’an, China, 2009: 506–509. doi: 10.1109/APSAR.2009.5374119.

LENG Xiangguang, JI Kefeng, YANG Kai, et al. A bilateral CFAR algorithm for ship detection in SAR images[J]. IEEE Geoscience and Remote Sensing Letters, 2015, 12(7): 1536–1540. doi: 10.1109/LGRS.2015.2412174

LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278–2324. doi: 10.1109/5.726791

HINTON G E and SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504–507. doi: 10.1126/science.1127647

KRIZHEVSKY A, SUTSKEVER I, and HINTON G E. ImageNet classification with deep convolutional neural networks[C]. The 25th International Conference on Neural Information Processing Systems, Lake Tahoe, USA, 2012: 1097–1105.

SIMONYAN K and ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv: 1409.1556, 2014.

SZEGEDY C, LIU Wei, JIA Yangqing, et al. Going deeper with convolutions[C]. 2015 IEEE Conference on Computer Vision and Pattern Recognition, Boston, USA, 2015: 1–9. doi: 10.1109/CVPR.2015.7298594.

HE Kaiming, ZHANG Xiangyu, REN Shaoqing, et al. Deep residual learning for image recognition[C]. 2016 IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, USA, 2016: 770–778. doi: 10.1109/CVPR.2016.90.

GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. 2014 IEEE Conference on Computer Vision and Pattern Recognition, Columbus, USA, 2014: 580–587. doi: 10.1109/CVPR.2014.81.

GIRSHICK R. Fast R-CNN[C]. 2015 IEEE International Conference on Computer Vision, Santiago, Chile, 2015: 1440–1448. doi: 10.1109/ICCV.2015.169.

REN Shaoqing, HE Kaiming, GIRSHICK R, et al. Faster R-CNN: Towards real-time object detection with region proposal networks[C]. The 28th International Conference on Neural Information Processing Systems, Montréal, Canada, 2015: 91–99.

杜蘭, 劉彬, 王燕, 等. 基于卷積神經(jīng)網(wǎng)絡(luò)的SAR圖像目標(biāo)檢測算法[J]. 電子與信息學(xué)報(bào), 2016, 38(12): 3018–3025. doi: 10.11999/JEIT161032

DU Lan, LIU Bin, WANG Yan, et al. Target detection method based on convolutional neural network for SAR image[J]. Journal of Electronics &Information Technology, 2016, 38(12): 3018–3025. doi: 10.11999/JEIT161032

ROSENBERG C, HEBERT M, and SCHNEIDERMAN H. Semi-supervised self-training of object detection models[C]. The 7th IEEE Workshops on Applications of Computer Vision, Breckenridge, USA, 2005: 29–36. doi: 10.1109/ACVMOT.2005.107.

ZHANG Fan, DU Bo, ZHANG Liangpei, et al. Weakly supervised learning based on coupled convolutional neural networks for aircraft detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(9): 5553–5563. doi: 10.1109/TGRS.2016.2569141

IOFFE S and SZEGEDY C. Batch normalization: Accelerating deep network training by reducing internal covariate shift[C]. The 32nd International Conference on Machine Learning, Lille, France, 2015: 448–456.

GLOROT X, BORDES A, and BENGIO Y. Deep sparse rectifier neural networks[C]. The 14th International Conference on Artificial Intelligence and Statistics, Fort Lauderdale, USA, 2011: 315–323.

GUTIERREZ D. MiniSAR: A review of 4-inch and 1-foot resolution Ku-band imagery[EB/OL]. https://www.sandia.gov/radar/Web/images/SAND2005-3706P-miniSAR-flight-SAR-images.pdf, 2005.

FARADSAR public Release Data[EB/OL]. https://www.sandia.gov/radar/complex_data/FARAD_KA_BAND.zip.

相關(guān)文章

施引文獻(xiàn)

資源附件(0)

訪問統(tǒng)計(jì)