在极化合成孔径雷达（PolSAR）地物分类研究中，基于卷积神经网络的图像分割算法存在高维特征信息冗余而导致的分类边界模糊、分类精度低、计算复杂等不足，提出一种基于卷积神经网络和EM算法的轻量化图像分割网络，称为低秩重构网络（low-rank-reconstruction-net，LRR-Net），应用于全极化SAR图像的地物分类。LRR-Net从极化目标分解的思想出发，利用EM算法对特征进行低秩重构，将特征从高维空间映射到低维空间，在减少参数的同时实现更精确的分类。用高分三号全极化图像数据对模型进行训练测试并评估，结果表明模型在保证分类精度的前提下，降低了模型复杂度。
In the study of the terrain classification based on the polarimetric synthetic aperture radar (PolSAR), the image segmentation algorithm based on deep neural network has the disadvantages of fuzzy classification boundary, low classification accuracy and complicated calculation caused by the redundancy of high-dimensional feature information. This paper proposes a lightweight segmentation network based on convolutional neural network and EM algorithm called low-rank-reconstruction-net (LRR-Net), which is applied to the terrain classification of fully PolSAR images. Starting from the idea of polarimetric target decomposition, LRR-Net uses the EM algorithm to perform low-rank reconstruction of features, maps the features from high-dimensional space to low-dimensional space, achieving higher classification accuracy while reducing parameters. The model is trained and evaluated in GF-3 fully PolSAR dataset, and the results show that the model complexity is reduced under the guarantee of the classification accuracy.
[1] Lee J S, Pottier E. 极化雷达成像基础与应用[M]. 洪文, 李洋, 尹嫱,等译. 北京:电子工业出版社, 2013.
[2] 赵昌锋. 基于深度学习的干涉SAR图像分类[D]. 西安:西安电子科技大学, 2015.
[3] 张澄波. 综合孔径雷达:原理、系统分析与应用[M]. 北京:科学出版社, 1989.
[4] Lee J S, Pottier E. Polarimetric radar imaging:from basics to applications[M]. Florida:CRC Press, 2009.
[5] Ye X, Zhang H, Wang C, et al. Classification of high-resolution SAR imagery by Random Forest classifier[C]//2013 Asia-Pacific Conference on Synthetic Aperture Radar(APSAR). September 23-27,2013,Tsukeba,Japan. IEEE, 2013:312-316.
[6] Demirhan M E, Salor Ö. Classification of targets in SAR images using SVM and k-NN techniques[C]//201624th Signal Processing and Communication Application Conference(SIU). May 16-19,2016,Zonguldak,Turkey. IEEE, 2016:1581-1584.DOI:10.1109/SIU.2016.7496056.
[7] Hou B, Yang C, Ren B, et al. Decomposition-feature-iterative-clustering-based superpixel segmentation for PolSAR image classification[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(8):1239-1243.DOI:10.1109/LGRS.2018.2833492.
[8] Deng L. Deep learning:methods and applications[J]. Foundations and Trends in Signal Processing, 2014, 7(3/4):197-387.DOI:10.1561/2000060039.
[9] Chen Y Q, Jiao L C, Li Y Y, et al. Multilayer projective dictionary pair learning and sparse autoencoder for PolSAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2017, 55(12):6683-6694.DOI:10.1109/TGRS.2017.2727067.
[10] Liu H Y, Yang S Y, Gou S P, et al. Polarimetric SAR feature extraction with neighborhood preservation-based deep learning[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2017, 10(4):1456-1466.DOI:10.1109/JSTARS.2016.2618891.
[11] Zhu H L, Lin N, Leung H, et al. Target classification from SAR imagery based on the pixel grayscale decline by graph convolutional neural network[J]. IEEE Sensors Letters, 2020, 4(6):1-4.DOI:10.1109/LSENS.2020.2995060.
[12] 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.
[13] Dempster A P, Laird N M, Rubin D B. Maximum likelihood from incomplete data via the EM algorithm[J]. Journal of the Royal Statistical Society:Series B (Methodological), 1977, 39(1):1-22.DOI:10.1111/j.2517-6161.1997.tb01600.x.
[14] Li X, Zhong Z S, Wu J L, et al. Expectation-maximization attention networks for semantic segmentation[C]//2019 IEEE/CVF International Conference on Computer Vision(ICCV). October 27-November 2,2019,Seoul,Korea(South). IEEE, 2019:9166-9175.DOI:10.1109/ICCV.2019.00926.
[15] He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition. June 27-30, 2016,Las Vegas,NV, USA. IEEE, 2016:770-778.DOI:10.1109/CVPR.2016.90.
[16] Hu J, Shen L, Albanie S. Squeeze-and-excitation networks[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. June 18-23, 2018, Salt Lake City,UT,USA. IEEE, 2018:7132-7141.DOI:10.1109/CVPR.2018.00745.
[17] 刘杰, 张庆君. 高分三号卫星及应用概况[J]. 卫星应用, 2018(6):12-16.DOI:10.3969/j.issn.1674-9030.2018.06.006.
[18] Long J, Shelhamer E, Darrell T. Fully convolutional networks for semantic segmentation[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition. June 7-12,2015,Boston,MA, USA. IEEE, 2015:3431-3440.DOI:10.1109/CVPR.2015.7298965.
[19] Zhao H S, Shi J P, Qi X J, et al. Pyramid scene parsing network[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. July 21-26, 2017,Honolulu,HI,USA. IEEE, 2017:6230-6239.DOI:10.1109/CVPR.2017.660.
[20] Chen L C, Papandreou G, Kokkinos I, et al. Deeplab:semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected CRFs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, 40(4):834-848.DOI:10.1109/TPAMI.2017.2699184.
