Please wait a minute...
 首页  期刊介绍 期刊订阅 联系我们 横山亮次奖 百年刊庆
 
最新录用  |  预出版  |  当期目录  |  过刊浏览  |  阅读排行  |  下载排行  |  引用排行  |  横山亮次奖  |  百年刊庆
清华大学学报(自然科学版)  2018, Vol. 58 Issue (2): 113-121    DOI: 10.16511/j.cnki.qhdxxb.2018.25.006
  计算机科学与技术 本期目录 | 过刊浏览 | 高级检索 |
基于泛化空间正则自动编码器的遥感图像识别
杨倩文, 孙富春
清华大学 计算机科学与技术系, 信息科学与技术国家实验室(筹), 智能技术与系统国家重点实验室, 北京 100084
Remote sensing image recognition based on generalized regularized auto-encoders
YANG Qianwen, SUN Fuchun
State Key Lab of Intelligent Technology and Systems, Tsinghua National Laboratory for Information Science and Technology, Department of Computer Science and Technology, Tsinghua University, Beijing 100084, China
全文: PDF(2250 KB)  
输出: BibTeX | EndNote (RIS)      
摘要 为了解决遥感图像中的合成孔径雷达(synthetic aperture radar,SAR)图像的样本稀缺问题,该文提出了针对这一小样本问题的泛化空间和泛化样本理论,将机器学习的分类问题转化为泛化空间中的样本密度估计问题。首先,通过研究泛化空间方法,针对有限样本的识别分类问题建立了样本密度估计模型,并从理论上验证了泛化空间方法的可行性;其次,在正则化自动编码器模型中,加入了泛化规则作为新的正则化因子对图像重构误差进行约束,针对有限样本问题建立泛化正则自动编码器(generalized auto-encoders,GAE),并提出利用该算法进行图像识别的模型;最后,将该模型应用于遥感图像小样本目标识别问题中。实验结果表明:GAE在SAR图像中具有最优的小样本学习能力,在样本数量有限的情况下,该方法表现出最小的重构误差和测试错误率。在小样本输入情况下,GAE模型实现了对MSTAR图像以及船舶SAR图像的识别分类,进一步证明了该算法相比于同类算法在SAR图像小样本识别问题中更具有优势。
服务
把本文推荐给朋友
加入引用管理器
E-mail Alert
RSS
作者相关文章
杨倩文
孙富春
关键词 泛化空间正则化自动编码器(GAE)合成孔径雷达(SAR)有限样本无监督学习    
Abstract:This paper addresses the problem of sample scarcity in SAR (synthetic aperture radar) images, which are very useful remote sensing sources but are very costly to obtain. SAR images have limited samples which makes it hard to get a standardized “good” model for even simple classification tasks. Then, the limited samples are expanded via generalization of the sample space using prior and a priori information including labels. Then, auto-learning models, regularized auto-encoders, are used to extract useful features from the generalized sample space. A regularization term is used to penalize generalized samples in an image recognition algorithm. The method produces good results on limit-sized subsets from the MNIST database and the SAR MSTAR image dataset. With limited input samples, the method gives better reconstruction errors and recognition accuracies in the SAR ship images than the state-of-art methods. Thus, this algorithm is shown to be successful in both theoretical analyses and tests.
Key wordsgeneralized sample space    generalized auto-encoders (GAE)    synthetic aperture radar (SAR)    limited samples    unsupervised learning
收稿日期: 2017-05-05      出版日期: 2018-02-15
ZTFLH:  TP753  
  TP391.4  
通讯作者: 孙富春,教授,E-mail:thusunfc@163.com     E-mail: thusunfc@163.com
引用本文:   
杨倩文, 孙富春. 基于泛化空间正则自动编码器的遥感图像识别[J]. 清华大学学报(自然科学版), 2018, 58(2): 113-121.
YANG Qianwen, SUN Fuchun. Remote sensing image recognition based on generalized regularized auto-encoders. Journal of Tsinghua University(Science and Technology), 2018, 58(2): 113-121.
链接本文:  
http://jst.tsinghuajournals.com/CN/10.16511/j.cnki.qhdxxb.2018.25.006  或          http://jst.tsinghuajournals.com/CN/Y2018/V58/I2/113
  图1 泛化空间M′上密度分布特性
  图2 泛化正则化样本的输入规则表示
  图3 J(x)在 MN I S T样本点邻域分布特性及其特征值分布
  图4 MS TAR重构误差与训练参数关系
  图5 测试错误率与训练样本个数关系
  表1 GAE泛化空间算法图像目标识别错误率
  图6 S AR 图像船舶目标原始图像
  图7 船舶图像数据集的分类实验曲线与结果
[1] KAPLAN L M. Analysis of multiplicative speckle models for template-based SAR ATR[J]. IEEE Transactions on Aerospace and Electronic Systems, 2001, 37(4):1424-1432.
[2] WANG J, SUN L J. Research on supervised manifold learning for SAR target classification[C]//IEEE International Conference on Computational Intelligence for Measurement Systems and Applications, CIMSA'09. Hong Kong, China:IEEE, 2009:140-142.
[3] BERISHA V, SHAH N, WAAGEN D, et al. Sparse manifold learning with applications to SAR image classification[C]//IEEE International Conference Acoustics, Speech and Signal Processing. Honolulu, HI, USA:IEEE, 2007:1086-1089.
[4] SCHERBAKOV A, HANSSEN R, VOSSELMAN G, et al. Ship wake detection using radon transforms of filtered SAR imagery[C]//Proceedings of the European Symposium on Satellite Remote Sensing:Microwave Sensing and Synthetic Aperture Radar. Taormina, Italy:International Society for Optics and Photonics, 1996:96-106.
[5] 种劲松, 朱敏慧. 高分辨率合成孔径雷达图像舰船检测方法[J]. 测试技术学报, 2003, 17(1):15-18.ZHONG J S, ZHU M H. Ship detection method of high-resolution SAR imagery[J]. J Test & Measurement Technology, 2003, 17(1):15-18. (in Chinese)
[6] CHAPELLE O, SCHLKOPF B, ZIEN A. Semi-supervised learning[J]. J Royal Stat Society, 2013, 49(2):215-239.
[7] 刘建伟, 刘媛, 罗雄麟. 半监督学习方法[J]. 计算机学报, 2015, 38(8):1592-1617.LIU J W, LIU Y, LUO X L. Semi-supervised learning methods[J]. Chinese J Computers, 2015, 38(8):1592-1617. (in Chinese)
[8] KINGMA D P, REZENDE D J, MOHAMED S, et al. Semi-supervised learning with deep generative models[J]. Advances in Neural Information Processing Systems, 2014, 4:3581-3589.
[9] SHINOZAKI T. Semi-supervised learning for convolutional neural networks using mild supervisory signals[C]//International Conference on Neural Information Processing. Kyoto, Japan:Springer International Publishing, 2016:381-388.
[10] PAPANDREOU G, CHEN L C, MURPHY K P, et al. Weakly and semi-supervised learning of a deep convolutional network for semantic image segmentation[C]//IEEE International Conference on Computer Vision. Santiago, Chile:IEEE, 2015:1742-1750.
[11] BLUM A. Semi-supervised learning[M]. New York:Springer, 2016:1936-1941.
[12] GUO G D, DYER C R. Learning from examples in the small sample case:Face recognition[J]. IEEE Trans on Systems Man & Cybernetics, 2005, 35(3):477-488.
[13] YAN H B, LU J W, ZHOU X Z, et al. Multi-feature multi-manifold learning for single-sample face recognition[J]. Neurocomputing, 2014, 143(16):134-143.
[14] LU J W, TAN Y P, WANG G. Discriminative multimanifold analysis for face recognition from a single training sample per person[C]//International Conference on Computer Vision. Barcelona, Spain:IEEE, 2011:1943-1950.
[15] TAN X Y, CHEN S C, ZHOU Z H, et al. Face recognition from a single image per person:A survey[J]. Pattern Recognition, 2006, 39(9):1725-1745.
[16] 陈维桓. 微分几何初步[M]. 北京:北京大学出版社, 1990.CHEN W H. Differential geometry preliminary[M]. Beijing:Peking University Press, 1990. (in Chinese)
[17] HUO X M, SMITH A K. A survey of manifold-based learning methods[C]//Recent Advances in Data Mining of Enterprise Data:Algorithms and Applications. Singapore:World Scientific Publishing Co. Inc., 2008:691-745.
[18] BENGIO Y, COURVILLE A, VINCENT P. Representation learning:A review and new perspectives[J]. IEEE Trans on Pattern Analysis & Machine Intelligence, 2013, 35(8):1798-1828.
[19] RIFAI S, VINCENT P, MULLER X, et al. Contractive auto-encoders:Explicit invariance during feature extraction[C]//International Conference on Machine Learning. Bellevue, WA, USA:Omnipress, 2011:833-840.
[20] ALAIN G, BENGIO Y. What regularized auto-encoders learn from the data-generating distribution[J]. J Machine Learning Research, 2014, 15(1):3563-3593.
[21] RIFAI S, MESNIL G, VINCENT P, et al. Higherorder contractive auto-encoder[C]//Joint European Conference on Machine Learning and Knowledge Discovery in Databases. Berlin, Heidelberg, Germany:Springer, 2011:645-660.
[22] YANG Q W, SUN F C. Small sample learning with high order contractive auto-encoders and application in SAR images[J]. arXiv preprint, 2017.
[23] 汪洪桥, 孙富春, 蔡艳宁, 等. 基于局部多分辨特征的SAR图像自动目标识别[J]. 清华大学学报(自然科学版), 2011, 51(8):1049-1054.WANG H Q, SUN F C, CAI Y N, et al. Automatic target recognition of SAR images based on local multi-resolution features[J]. J Tsinghua University (Science and Technology), 2011, 51(8):1049-1054.(in Chinese)
[1] 王素格, 李大宇, 李旸. 基于联合模型的商品口碑数据情感挖掘[J]. 清华大学学报(自然科学版), 2017, 57(9): 926-931.
[2] 李增辉, 常雯, 杨健. 基于外推陷波滤波的孤立强散射旁瓣抑制[J]. 清华大学学报(自然科学版), 2015, 55(5): 503-507.
[3] 马文婷,杨健,崔一,高伟. 地面高度起伏对SAR图像匹配性能影响的定量分析[J]. 清华大学学报(自然科学版), 2014, 54(4): 432-436.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
版权所有 © 《清华大学学报(自然科学版)》编辑部
本系统由北京玛格泰克科技发展有限公司设计开发 技术支持:support@magtech.com.cn