基于超像素与LightGBM的极化SAR图像地物分类

doi:10.7523/j.ucas.2022.023

摘要/Abstract

摘要： 极化合成孔径雷达(SAR)图像的相干斑噪声降低了地物分类的准确率;联合极化SAR图像众多特征分类,过大的输入特征维度导致分类耗时长。为解决上述问题,提出一种基于超像素与LightGBM的分类算法。该算法充分利用极化SAR图像的极化特征与纹理特征,具备较强的分类能力;采用LightGBM处理大维度输入特征,能够快速得到基于像素的初级分类结果;利用SLIC生成基于超像素的极化SAR图像,并在各超像素内逐像素投票得到基于超像素的二级分类结果,抑制了相干斑的影响。利用实测极化SAR数据进行实验,分类的总体准确率超过97%,且分类耗时短。

关键词: 极化合成孔径雷达, 地物分类, 超像素, SLIC, LightGBM

Abstract: Speckle noise of polarimetric synthetic aperture radar(SAR) image reduces the accuracy of terrain classification. Combine multiple features of polarimetric SAR image to do classification, and the large dimension of input features consumes too much time. To handle the above problems, we propose a classification algorithm based on superpixel and LightGBM. With polarimetric features and texture features, the algorithm is good at classification. LightGBM is used to deal with large dimension of input features, which can obtain the pixel-based first-level classification result efficiently. SLIC is used to generate the superpixel-based polarimetric SAR image, and the superpixel-based two-level classification result is obtained by voting pixel by pixel in each superpixel, which solves the influence of speckle noise. Experimental results, using the measured polarimetric SAR data, show that the overall classification accuracy is more than 97%, and it has a low time-consuming.

Key words: polarimetric SAR, terrain classification, superpixel, SLIC, LightGBM

中图分类号:

TP753

王懿泽, 孙吉利, 闫成杰, 张政. 基于超像素与LightGBM的极化SAR图像地物分类[J]. 中国科学院大学学报, 2023, 40(5): 658-669.

WANG Yize, SUN Jili, YAN Chengjie, ZHANG Zheng. Polarimetric SAR image terrain classification based on superpixel and LightGBM[J]. Journal of University of Chinese Academy of Sciences, 2023, 40(5): 658-669.

参考文献

[1] 张妙然, 刘畅. 基于特征筛选和二级分类的极化SAR建筑提取算法[J]. 中国科学院大学学报, 2018, 35(1):89-95. DOI:10.7523/j.issn.2095-6134.2018.01.012.
[2] 马肖肖, 程博, 刘岳明, 等. 基于极化特征和纹理特征的PolSAR影像建筑物提取方法[J]. 中国科学院大学学报, 2019, 36(5):682-693. DOI:10.7523/j.issn.2095-6134.2019.05.014.
[3] 刘杉, 张风丽, 韦诗莹, 等. 基于极化分解组合的SAR图像视觉优化和建筑物损毁评估[J]. 中国科学院大学学报, 2020, 37(6):750-759. DOI:10.7523/j.issn.2095-6134.2020.06.005.
[4] Lee J S, Grunes M R, 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.
[5] Gomez L, Alvarez L, Mazorra L, et al. Fully PolSAR image classification using machine learning techniques and reaction-diffusion systems[J]. Neurocomputing, 2017, 255:52-60. DOI:10.1016/j.neucom.2016.08.140.
[6] Wu Q, Hou B, Wen Z D, et al. Variational learning of mixture wishart model for PolSAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(1):141-154. DOI:10.1109/TGRS.2018.2852633.
[7] Bi H X, Xu L, Cao X Y, et al. Polarimetric SAR image semantic segmentation with 3D discrete wavelet transform and Markov random field[J]. IEEE Transactions on Image Processing, 2020, 29:6601-6614. DOI:10.1109/TIP.2020.2992177.
[8] Song W Y, Li M, Zhang P, et al. Mixture WGГ-MRF model for PolSAR image classification[J]. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(2):905-920. DOI:10.1109/TGRS.2017.2756621.
[9] Liu C, Liao W Z, Li H C, et al. Semi-supervised classification of polarimetric SAR images using Markov random field and two-level wishart mixture model[C]//IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium. July 28-August 2, 2019, Yokohama, Japan. IEEE, 2019:990-993. DOI:10.1109/IGARSS.2019.8898985.
[10] Guan D D, Xiang D L, Dong G G, et al. SAR image classification by exploiting adaptive contextual information and composite kernels[J]. IEEE Geoscience and Remote Sensing Letters, 2018, 15(7):1035-1039. DOI:10.1109/LGRS.2018.2821711.
[11] Li Y, Yin Q, Wang Y P, et al. Multi-aspect polarimetric SAR image scattering feature information coding and classification with machine learning approach[C]//EUSAR 2021; 13th European Conference on Synthetic Aperture Radar. March 29-April 1, 2021, online. VDE, 2021:1-4. DOI:10.1109/APSAR46974.2019.9048589.
[12] Yin Q, Cheng J D, Zhang F, et al. Interpretable POLSAR image classification based on adaptive-dimension feature space decision tree[J]. IEEE Access, 2020, 8:173826-173837. DOI:10.1109/ACCESS.2020.3023134.
[13] Zhang X T, Xu J, Chen Y Y, et al. Coastal wetland classification with GF-3 polarimetric SAR imagery by using object-oriented random forest algorithm[J]. Sensors, 2021, 21(10):3395. DOI:10.3390/s21103395.
[14] Du P J, Samat A, Waske B, et al. Random Forest and Rotation Forest for fully polarized SAR image classification using polarimetric and spatial features[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 2015, 105:38-53. DOI:10.1016/j.isprsjprs.2015.03.002.
[15] Ustuner M, Sanli F B, Abdikan S, et al. A booster analysis of extreme gradient boosting for crop classification using PolSAR imagery[C]//2019 8th International Conference on Agro-Geoinformatics (Agro-Geoinformatics). July 16-19, 2019, Istanbul, Turkey. IEEE, 2019:1-4. DOI:10.1109/Agro-Geoinformatics.2019.8820698.
[16] Haralick R M, Shanmugam K, Dinstein I. Textural features for image classification[J]. IEEE Transactions on Systems, Man, and Cybernetics, 1973, SMC-3(6):610-621. DOI:10.1109/TSMC.1973.4309314.
[17] 邓滢, 张红, 王超, 等. 结合纹理与极化分解的面向对象极化SAR水体提取方法[J]. 遥感技术与应用, 2016, 31(4):714-723. DOI:10.11873/j.issn.1004-0323.2016.4.0714.
[18] Zhai W, Shen H F, Huang C L, et al. Fusion of polarimetric and texture information for urban building extraction from fully polarimetric SAR imagery[J]. Remote Sensing Letters, 2016, 7(1):31-40. DOI:10.1080/2150704X.2015.1101179.
[19] Nie Y L, Zeng Q M, Zhang H Z, et al. Building damage detection based on OPCE matching algorithm using a single post-event PolSAR data[J]. Remote Sensing, 2021, 13(6):1146. DOI:10.3390/rs13061146.
[20] Masjedi A, Zoej M J V, Maghsoudi Y. Classification of polarimetric SAR images based on modeling contextual information and using texture features[J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(2):932-943. DOI:10.1109/TGRS.2015.2469691.
[21] Luo S Y, Sarabandi K, Tong L, et al. A SAR image classification algorithm based on multi-feature polarimetric parameters using FOA and LS-SVM[J]. IEEE Access, 2019, 7:175259-175276. DOI:10.1109/ACCESS.2019.2957547.
[22] Pierce L E, Ulaby F T, Sarabandi K, et al. Knowledge-based classification of polarimetric SAR images[J]. IEEE Transactions on Geoscience and Remote Sensing, 1994, 32(5):1081-1086. DOI:10.1109/36.312896.
[23] Cloude S R, 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.
[24] Kandaswamy U, Adjeroh D A, Lee M C. Efficient texture analysis of SAR imagery[J]. IEEE Transactions on Geoscience and Remote Sensing, 2005, 43(9):2075-2083. DOI:10.1109/TGRS.2005.852768.
[25] Yin Q, Hong W, Zhang F, et al. Optimal combination of polarimetric features for vegetation classification in PolSAR image[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2019, 12(10):3919-3931. DOI:10.1109/JSTARS.2019.2940973.
[26] Ke G, Meng Q, Finley T, et al. Lightgbm:a highly efficient gradient boosting decision tree[C/OL]. 31st Conference on Neural Information Processing Systems (NIPS 2017). December 4-9, 2017, Long Beach, CA, USA. Curran Associates, Inc, 2017.[2022-01-15]. https://proceedings.neurips.cc/paper/2017/file/6449f44a102fde848669bdd9eb6b76fa-Paper.pdf.
[27] 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.
[28] 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.
[29] Xie W, Ma G N, Zhao F, et al. PolSAR image classification via a novel semi-supervised recurrent complex-valued convolution neural network[J]. Neurocomputing, 2020, 388:255-268. DOI:10.1016/j.neucom.2020.01.020.
[30] Zhang P, Tan X F, Li B B, et al. PolSAR image classification using hybrid conditional random fields model based on complex-valued 3-D CNN[J]. IEEE Transactions on Aerospace and Electronic Systems, 2021, 57(3):1713-1730. DOI:10.1109/TAES.2021.3050648.
[31] Xiao D L, Liu C, Wang Q, et al. PolSAR image classification based on dilated convolution and pixel-refining parallel mapping network in the complex domain[EB/OL]. arXiv:1909.10783v2. (2019-09-24)[2022-03-15]. https://arxiv.org/abs/1909.10783v2.
[32] Dong H W, Zhang L M, Lu D, et al. Attention-based polarimetric feature selection convolutional network for PolSAR image classification[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19:1-5. DOI:10.1109/LGRS.2020.3021373.
[33] Cheng J D, Zhang F, Xiang D L, et al. PolSAR image land cover classification based on hierarchical capsule network[J]. Remote Sensing, 2021, 13(16):3132. DOI:10.3390/rs13163132.
[34] Ren X F, Malik J. Learning a classification model for segmentation[C]//Proceedings Ninth IEEE International Conference on Computer Vision. October 13-16, 2003, Nice, France. IEEE, 2003:10-17. DOI:10.1109/ICCV.2003.1238308.
[35] Yan J J, Yu Y N, Zhu X Y, et al. Object detection by labeling superpixels[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition. June 7-12, 2015, Boston, MA, USA. IEEE, 2015:5107-5116. DOI:10.1109/CVPR.2015.7299146.
[36] Yeo D, Son J, Han B, et al. Superpixel-based tracking-by-segmentation using Markov chains[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition. July 21-26, 2017, Honolulu, HI, USA. IEEE, 2017:511-520. DOI:10.1109/CVPR.2017.62.
[37] Sun W, Liao Q M, Xue J H, et al. SPSIM:a superpixel-based similarity index for full-reference image quality assessment[J]. IEEE Transactions on Image Processing:a Publication of the IEEE Signal Processing Society, 2018, 27(9):4232-4244. DOI:10.1109/TIP.2018.2837341.
[38] Achanta R, Shaji A, Smith K, et al. SLIC superpixels compared to state-of-the-art superpixel methods[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(11):2274-2282. DOI:10.1109/TPAMI.2012.120.
[39] Friedman J H. Greedy function approximation:a gradient boosting machine[J]. The Annals of Statistics, 2001, 29(5):1189-1232. DOI:10.1214/aos/1013203451.
[40] Zuo Y X, Guo J Y, Zhang Y T, et al. A deep vector quantization clustering method for polarimetric SAR images[J]. Remote Sensing, 2021, 13(11):2127. DOI:10.3390/rs13112127.