基于深度卷积神经网络和条件随机场模型的PolSAR图像地物分类方法

胡涛 李卫华 秦先祥 王鹏 余旺盛 李军

胡涛, 李卫华, 秦先祥, 等. 基于深度卷积神经网络和条件随机场模型的PolSAR图像地物分类方法[J]. 雷达学报, 2019, 8(4): 471–478. doi: 10.12000/JR18065
引用本文: 胡涛, 李卫华, 秦先祥, 等. 基于深度卷积神经网络和条件随机场模型的PolSAR图像地物分类方法[J]. 雷达学报, 2019, 8(4): 471–478. doi: 10.12000/JR18065
HU Tao, LI Weihua, QIN Xianxiang, et al. Terrain classification of polarimetric synthetic aperture radar images based on deep learning and conditional random field model[J]. Journal of Radars, 2019, 8(4): 471–478. doi: 10.12000/JR18065
Citation: HU Tao, LI Weihua, QIN Xianxiang, et al. Terrain classification of polarimetric synthetic aperture radar images based on deep learning and conditional random field model[J]. Journal of Radars, 2019, 8(4): 471–478. doi: 10.12000/JR18065

基于深度卷积神经网络和条件随机场模型的PolSAR图像地物分类方法

doi: 10.12000/JR18065
基金项目: 国家自然科学基金(41601436, 61403414, 61703423),陕西省自然科学基础研究计划(2018JM4029)
详细信息
    作者简介:

    胡涛:胡 涛(1994–),男,湖南浏阳人,空军工程大学信息与导航学院硕士,研究方向为计算机视觉。E-mail: 1862965@163.com

    李卫华(1964–),男,空军工程大学信息与导航学院教授,研究方向为指挥信息系统。E-mail: lwh_kgd@163.com 

    秦先祥(1986–),男,广西阳朔人,空军工程大学信息与导航学院讲师,研究方向为SAR图像处理与分析。E-mail: qinxianxiang@126.com

    王鹏:王 鹏(1985–),男,空军工程大学信息与导航学院副教授,硕士生导师,研究方向为信息融合处理与分布式协同控制。E-mail: wangpeng@163.com

    余旺盛(1985–),男,湖南平江人,空军工程大学信息与导航学院讲师,研究方向为计算机视觉与图像处理。E-mail: 853994682@qq.com

    李军:李 军(1983–),男,湖南邵阳人,空军工程大学信息与导航学院讲师,研究方向为信息处理技术。E-mail: 108857769@qq.com

    通讯作者:

    秦先祥   qinxianxiang@126.com

  • 中图分类号: TP391

Terrain Classification of Polarimetric Synthetic Aperture Radar Images Based on Deep Learning and Conditional Random Field Model

Funds: The National Natural Science Foundation of China (41601436, 61403414, 61703423), The Natural Science Foundation Research Project of Shaanxi Province (2018JM4029)
More Information
  • 摘要: 近年来,极化合成孔径雷达(PolSAR)图像地物分类得到了深入研究。传统的PolSAR图像地物分类方法采用的特征往往需要针对具体问题进行设计,特征表征性不强。因此,该文提出一种基于卷积神经网络(CNN)和条件随机场(CRF)模型的PolSAR图像地物分类方法。利用预训练好的实现图像分类任务的卷积神经网络模型(VGG-Net-16)提取表征能力更强的图像特征,再通过CRF模型对多特征及上下文信息的有效利用来实现图像的地物分类。实验结果表明,与3种利用传统经典特征的方法相比,该方法能够提取更有效的特征,取得了更高的总体分类精度和Kappa系数。

     

  • 图  1  深度CRF模型流程图

    Figure  1.  The flow chart of deep CRF model

    图  2  Flevoland数据分类结果对比图

    Figure  2.  Comparison of Flevoland data classification results

    图  3  Oberpfaffenhofen数据分类结果对比图

    Figure  3.  Comparison of Oberpfaffenhofendata classification results

    图  4  不同层特征分类精度对比图

    Figure  4.  Accuracy comparison results of different layer classification results

    表  1  传统方法中用到的特征

    Table  1.   The features used in the traditional methods

    Cloude分解Freeman分解协方差矩阵对角线
    $H,\alpha ,A,{\lambda _{1}},{\lambda _{{2}}},{\lambda _{{3}}}$Ps, Pd, PvC11, C22, C33
    下载: 导出CSV

    表  2  Flevoland数据分类精度

    Table  2.   The classification accuracy of Flevoland data

    类别方法1方法2方法3方法4方法5本文方法
    豆类0.9710.8330.9670.8630.9200.808
    森林0.7590.9400.7330.9430.9450.868
    土豆0.6800.8400.8210.5780.8720.808
    苜蓿0.6090.8920.7190.7810.9320.990
    小麦0.9340.8810.8640.7920.9360.981
    裸地0.5140.8710.9030.9800.9980.899
    甜菜0.9130.9030.8950.9050.8970.978
    油菜籽0.5720.7820.6270.7580.9340.964
    豌豆0.5890.8210.8200.8010.9010.854
    草地0.9620.7740.8380.9120.8020.968
    水体0.7010.9700.5260.7030.9880.888
    总精度0.7510.8700.7780.7970.9330.905
    Kappa系数0.7200.8540.7520.7740.9110.890
    训练(s)798771877121170661052
    测试(s)2.92.73.04.18.43.8
    下载: 导出CSV

    表  3  Oberpfaffenhofen数据分类精度

    Table  3.   The classification accuracy of Oberpfaffenhofen data

    类别方法1方法2方法3本文方法
    建筑区域0.6960.6450.7120.903
    林地0.8950.8960.7000.777
    开放区域0.6220.8430.8740.947
    总精度0.6910.8040.8000.903
    Kappa系数0.5290.6800.6680.834
    下载: 导出CSV
  • [1] NOVAK L M and BURL M C. Optimal speckle reduction in polarimetric SAR imagery[J]. IEEE Transactions on Aerospace and Electronic Systems, 1990, 26(2): 293–305. doi: 10.1109/7.53442
    [2] RANSON K J, SUN G, WEISHAMPEL J F, et al. An evaluation of AIRSAR and SIR-C/XSAR images for northern forest ecological studies in Maine, USA[C]. Proceedings of 1995 International Geoscience and Remote Sensing Symposium, IGARSS’95. Quantitative Remote Sensing for Science and Applications, Firenze, Italy, 1995: 994–996. doi: 10.1109/IGARSS.1995.521118.
    [3] YANG W, ZHANG X, CHEN L J, et al. Semantic segmentation of polarimetric SAR imagery using conditional random fields[C]. Proceedings of 2010 IEEE International Geoscience and Remote Sensing Symposium, Honolulu, HI, USA, 2010: 1593–1596. doi: 10.1109/IGARSS.2010.5652378.
    [4] CLOUDE S R and POTTIER E. An entropy based classification scheme for land applications of polarimetric SAR[J]. IEEE Transactions on Geoscience and Remote Sensing, 1997, 35(1): 68–78. doi: 10.1109/36.551935
    [5] ZHAO L W, ZHOU X G, JIANG Y M, et al. Iterative classification of polarimetric SAR image based on the freeman decomposition and scattering entropy[C]. Proceedings of the 1st Asian and Pacific Conference on Synthetic Aperture Radar, Huangshan, China, 2007: 473–476. doi: 10.1109/APSAR.2007.4418653.
    [6] LEE J S, GRUNES M R, and KWOK R. Classification of multi-look polarimetric SAR imagery based on complex wishart distribution[J]. International Journal of Remote Sensing, 1994, 15(11): 2299–2311. doi: 10.1080/01431169408954244
    [7] BEAULIEU J M and TOUZI R. Segmentation of textured polarimetric SAR scenes by likelihood approximation[J]. IEEE Transactions on Geoscience and Remote Sensing, 2004, 42(10): 2063–2072. doi: 10.1109/tgrs.2004.835302
    [8] WU Y H, JI K F, YU W X, et al. Region-based classification of polarimetric SAR images using wishart MRF[J]. IEEE Geoscience and Remote Sensing Letters, 2008, 5(4): 668–672. doi: 10.1109/LGRS.2008.2002263
    [9] 周晓光, 匡纲要, 万建伟. 极化SAR图像分类综述[J]. 信号处理, 2008, 24(5): 806–812. doi: 10.3969/j.issn.1003-0530.2008.05.023

    ZHOU Xiao-guang, KUANG Gang-yao, and WAN Jian-wei. A review of polarimetric SAR image classification[J]. Signal Processing, 2008, 24(5): 806–812. doi: 10.3969/j.issn.1003-0530.2008.05.023
    [10] 胡涛, 李卫华, 秦先祥, 等. 基于深度CRF模型的图像语义分割方法[J]. 空军工程大学学报(自然科学版), 2018, 19(5): 52–57.

    HU Tao, LI Wei-hua, QIN Xian-xiang, et al. An image semantic segmentation based on deep CRF model[J]. Journal of Air Force Engineering University (Natural Science Edition), 2018, 19(5): 52–57.
    [11] XIE W, JIAO L C, HOU B, et al. POLSAR image classification via wishart-AE model or wishart-CAE model[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017, 10(8): 3604–3615. doi: 10.1109/JSTARS.2017.2698076
    [12] ZHAO Z Q, JIAO L C, ZHAO J Q, et al. Discriminant deep belief network for high-resolution SAR image classification[J]. Pattern Recognition, 2017, 61: 686–701. doi: 10.1016/j.patcog.2016.05.028
    [13] GAO F, HUANG T, WANG J, et al. Dual-branch deep convolution neural network for polarimetric SAR image classification[J]. Applied Sciences, 2017, 7(5): 447. doi: 10.3390/app7050447
    [14] ZHOU Y, WANG H P, XU F, et al. Polarimetric SAR image classification using deep convolutional neural networks[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(12): 1935–1939. doi: 10.1109/LGRS.2016.2618840
    [15] ZHANG Z M, WANG H P, XU F, 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. doi: 10.1109/TGRS.2017.2743222
    [16] WANG L, XU X, DONG H, et al. Multi-pixel simultaneous classification of PolSAR image using convolutional neural networks[J]. Sensors, 2018, 18(3): 769. doi: 10.3390/s18030769
    [17] KRIZHEVSKY A, SUTSKEVER I, and HINTON G E. ImageNet classification with deep convolutional neural networks[C]. Proceedings of the 25th International Conference on Neural Information Processing Systems, Lake Tahoe, Nevada, USA, 2012: 1097–1105. doi: 10.1145/3065386.
    [18] SZEGEDY C, LIU W, JIA Y Q, et al.. Going deeper with convolutions[C]. Proceedings of 2015 IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 2015: 1–9. doi: 10.1109/CVPR.2015.7298594.
    [19] SIMONYAN K and ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556, 2014.
    [20] RAZAVIAN A S, AZIZPOUR H, SULLIVAN J, et al. CNN features off-the-shelf: An astounding baseline for recognition[C]. Proceedings of 2014 IEEE Conference on Computer Vision and Pattern Recognition Workshops, Columbus, OH, USA, 2014: 512–519. doi: 10.1109/CVPRW.2014.131.
    [21] MIKA S, SCHÖLKOPF B, SMOLA A, et al.. Kernel PCA and de-noising in feature spaces[C]. Proceedings of 1998 Conference on Advances in Neural Information Processing Systems II, Cambridge, MA, USA, 1999: 536–542.
    [22] LAFFERTY J D, MCCALLUM A, and PEREIRA F C N. Conditional random fields: Probabilistic models for segmenting and labeling sequence data[C]. Proceedings of the 18th International Conference on Machine Learning, San Francisco, CA, USA, 2001: 282–289.
    [23] LI S Z. Markov Random Field Modeling in Computer Vision[M]. New York: Springer, 1995.
    [24] PLATT J C. Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods[C]. Proceedings of the Advances in large Margin Classifiers, Cambrige, MA, USA, 1999: 61–74.
    [25] DOMKE J. Learning graphical model parameters with approximate marginal inference[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(10): 2454–2467. doi: 10.1109/TPAMI.2013.31
    [26] VEDALDI A and LENC K. MatConvNet: Convolutional neural networks for MATLAB[C]. Proceedings of the 23rd ACM International Conference on Multimedia, Brisbane, Australia, 2015: 689–692. doi: 10.1145/2733373.2807412.
    [27] 韩萍, 韩宾宾. 基于典型散射差异指数的PolSAR图像Lee滤波[J]. 系统工程与电子技术, 2018, 40(2): 287–294.

    HAN Ping and HAN Bin-bin. Lee filter of PolSAR image based on typical scattering difference index[J]. Systems Engineering and Electronics, 2018, 40(2): 287–294.
  • 加载中
图(4) / 表(3)
计量
  • 文章访问数:  2761
  • HTML全文浏览量:  596
  • PDF下载量:  280
  • 被引次数: 0
出版历程
  • 收稿日期:  2018-08-31
  • 修回日期:  2018-12-26
  • 刊出日期:  2019-08-28

目录

    /

    返回文章
    返回