考虑空间相关性的半监督局部保持投影的高光谱图像特征提取
|
摘要
由于高光谱图像具有波段多、波段间信息冗余、空间信息相关等特点,提出一种考虑空间相关性的半监督局部保持投影(LPP)的高光谱图像特征提取算法(LPP-SCSSFE)。该算法利用保存高光谱图像空间近邻结构的空间距离,及保持图像光谱相似性的类内判别权值和类间判别权值,定义新的同物异谱、同谱异物像元权值计算函数,结合LPP提取高光谱图像特征,从而最大化同类地物间的相似性和异类地物间的差异性。在Indian Pines和Pavia University两个数据集上,通过高光谱图像分类实验对本文提出的LPP-SCSSFE算法进行验证,算法最高总体分类精度分别达到87.50%和91.29%,优于现有的特征提取算法。结果表明,本文算法充分考虑高光谱图像的空间相关性和光谱相似性,能够有效提取出有代表性的特征,提高分类精度。
Based on the characteristics of multi-band,inter-band information redundancy and spatial information correlation of hyperspectral images,a spatially-correlated and semi-supervised feature extraction(SCSSFE)algorithm with locality preserving projection(LPP)is proposed.This algorithm defines a new pixel weight calculation function for the different spectral characteristics with the same objects and the different objects with the same spectral characteristics to preserve the spatial distance and the spectral similarity of hyperspectral image by means of the neighbor structure in image space and the intra-class and inter-class discriminant weights.Then,the features of hyperspectral images are extracted by the weight function combined with LPP.Thus the similarity among the same objects and the discrepancy among the different objects are maximized.The proposed LPP-SCSSFE algorithm is verified through the hyperspectral image classification experiments on the two datasets of Indian Pines and Pavia University.The highest overall classification accuracies of the LPP-SCSSFE algorithm reach 87.50%and 91.29%for the respective datasets,better than those of the existing feature extraction algorithms.These results indicate that the spatial correlation and the spectral similarity of hyperspectral images are fully taken into account in the proposed algorithm,and thus the more representative features are extracted and the classification accuracy is enhanced.
Based on the characteristics of multi-band,inter-band information redundancy and spatial information correlation of hyperspectral images,a spatially-correlated and semi-supervised feature extraction(SCSSFE)algorithm with locality preserving projection(LPP)is proposed.This algorithm defines a new pixel weight calculation function for the different spectral characteristics with the same objects and the different objects with the same spectral characteristics to preserve the spatial distance and the spectral similarity of hyperspectral image by means of the neighbor structure in image space and the intra-class and inter-class discriminant weights.Then,the features of hyperspectral images are extracted by the weight function combined with LPP.Thus the similarity among the same objects and the discrepancy among the different objects are maximized.The proposed LPP-SCSSFE algorithm is verified through the hyperspectral image classification experiments on the two datasets of Indian Pines and Pavia University.The highest overall classification accuracies of the LPP-SCSSFE algorithm reach 87.50%and 91.29%for the respective datasets,better than those of the existing feature extraction algorithms.These results indicate that the spatial correlation and the spectral similarity of hyperspectral images are fully taken into account in the proposed algorithm,and thus the more representative features are extracted and the classification accuracy is enhanced.
引文
[1] Liu B,Yu X C,Zhang P Q,et al.Supervised deepfeature extraction for hyperspectral imageclassification[J].IEEE Transactions on Geoscienceand Remote Sensing,2018,56(4):1909-1921.
[2] Wang J N.Hyperspectral image classification basedon joint sparse representation and morphologicalfeature extraction[J].Laser &OptoelectronicsProgress,2016,53(8):082801.王佳宁.基于联合稀疏表示与形态特征提取的高光谱图像分类[J].激光与光电子学进展,2016,53(8):082801.
[3] Liao J S, Wang L G. Hyperspectral imageclassification method based on fusion with two kindsof spatial information[J].Laser &OptoelectronicsProgress,2017,54(8):081002.廖建尚,王立国.两类空间信息融合的高光谱图像分类方法[J].激光与光电子学进展,2017,54(8):081002.
[4] Huang H,He K,Zheng X L,et al.Spatial-spectralfeature extraction of hyperspectral image based ondeep learning[J].Laser &Optoelectronics Progress,2017,54(10):101001.黄鸿,何凯,郑新磊,等.基于深度学习的高光谱图像空-谱联合特征提取[J].激光与光电子学进展,2017,54(10):101001.
[5] Hou B H,Yao M L,Wang R,et al.Spatial-spectralsemi-supervised local discriminant analysis forhyperspectral image classification[J].Acta OpticaSinica,2017,37(7):0728002.侯榜焕,姚敏立,王榕,等.面向高光谱图像分类的空谱半监督局部判别分析[J].光学学报,2017,37(7):0728002.
[6] Li D R,Zhang L P,Xia G S.Automatic analysisandmining of remote sensing big data[J].ActaGeodaetica et Cartographica Sinica,2014,43(12):1211-1216.李德仁,张良培,夏桂松.遥感大数据自动分析与数据挖掘[J].测绘学报,2014,43(12):1211-1216.
[7] Lu G F,Wang Y,Zou J,et al.Matrix exponentialbased discriminant locality preserving projections forfeature extraction[J].Neural Networks,2018,97:127-136.
[8] Sellami A,Farah I R.A spatial hypergraph basedsemi-supervised band selection method forhyperspectral imagery semantic interpretation[J].World Academy of Science, Engineering andTechnology,International Journal of Computer,Electrical, Automation, Control and InformationEngineering,2016,10(10):1839-1846.
[9] Zhang B. Advancement of hyperspectral imageprocessing and information extraction[J].Journal ofRemote Sensing,2016,20(5):1062-1090.张兵.高光谱图像处理与信息提取前沿[J].遥感学报,2016,20(5):1062-1090.
[10] Gan Y H,Luo F L,Liu J H,et al.Featureextraction based multi-structure manifold embeddingfor hyperspectral remote sensing image classification[J].IEEE Access,2017,5:25069-25080.
[11] Lodha S P,Kamlapur S.Dimensionality reductiontechniques for hyperspectral images[J].InternationalJournal of Application or Innovation in Engineering &Management(IJAIEM),2014,3(10):92-99.
[12] Zhang X R, Huyan N,Zhou N,et al.Semi-supervised sparse dimensionality reduction forhyperspectral image classification[C]∥2016IEEERegion 10 Conference, November 22-25,2016,Singapore.New York:IEEE,2016:2830-2833.
[13] Xue S Z,Tan R,Chen X H.Method of kernel-basedsemi-supervised locality preserving projection[J].Journal of Computer Applications,2012,32(8):2235-2237,2244.薛寺中,谈锐,陈秀宏.基于核的半监督的局部保留投影降维方法[J].计算机应用,2012,32(8):2235-2237,2244.
[14] Wold S,Esbensen K,Geladi P.Principal componentanalysis[J].Chemometrics and Intelligent LaboratorySystems,1987,2(1/2/3):37-52.
[15] He X F.Locality preserving projectionsp[D].Chicago:University of Chicago,2005.
[16] Zhang J B, Wang J K,Cai X.Sparse localitypreserving discriminative projections for facerecognition[J].Neurocomputing,2017,260:321-330.
[17] Wan M H,Lai Z H.Multi-manifold locality graphembedding based on the maximum margin criterion(MLGE/MMC)for face recognition[J].IEEEAccess,2017,5:9823-9830.
[18] Huang Y, Tang K,Sun Z.Locality sensitivediscriminant analysis for classification ofhyperspectral data[C]∥2014 7th InternationalCongress on Image and Signal Processing,October14-16,2014,Dalian,China.New York:IEEE,2014:292-296.
[19] He X F,Cai D,Yan S C,et al.Neighborhoodpreserving embedding[C]∥Tenth IEEE InternationalConference on Computer Vision(ICCV’05),October17-21,2005,Beijing,China.New York:IEEE,2005,2:1208-1213.
[20] Wu Z Y,Hou B Y,Wang H B,et al.Semi-supervised sparsity neighboring preserving projection[J].Systems Engineering and Electronics,2018,40(4):934-940.吴振宇,侯冰洋,王辉兵,等.半监督稀疏近邻保持投影[J].系统工程与电子技术,2018,40(4):934-940.
[21] Lu X H,He Z Y,Yi S Y,eta l.Joint of locality-andglobality-preserving projections[J].Signal,Imageand Video Processing,2018,12(3):565-572.
[22] Pan Y S,Wang P F,Huang H,eta l.Hyperspectralremote sensing image classification based on SSLPP[J].Computer Science,2013(S2):333-336,373.潘银松,王攀峰,黄鸿,等.基于SSLPP算法对高光谱遥感影像分类[J].计算机科学,2013(S2):333-336,373.
[23] Du W B,LüM,Hou Q L,et al.Semisuperviseddimension reduction based on pairwise constraintpropagation for hyperspectral images[J].IEEEGeoscience and Remote Sensing Letters,2016,13(12):1880-1884.
[24] Huang H, Huang Y B.Improved discriminantsparsity neighborhood preserving embedding forhyperspectral image classification[J].Neurocomputing,2014,136:224-234.
[25] He F,Wang R,Yu Q,et al.Feature extraction ofhyperspectral images of weighted spatial and spectrallocality preserving projection(WSSLPP)[J].Opticsand Precision Engineering,2017,25(1):263-273.何芳,王榕,于强,等.加权空谱局部保持投影的高光谱图像特征提取[J].光学精密工程,2017,25(1):263-273.
[2] Wang J N.Hyperspectral image classification basedon joint sparse representation and morphologicalfeature extraction[J].Laser &OptoelectronicsProgress,2016,53(8):082801.王佳宁.基于联合稀疏表示与形态特征提取的高光谱图像分类[J].激光与光电子学进展,2016,53(8):082801.
[3] Liao J S, Wang L G. Hyperspectral imageclassification method based on fusion with two kindsof spatial information[J].Laser &OptoelectronicsProgress,2017,54(8):081002.廖建尚,王立国.两类空间信息融合的高光谱图像分类方法[J].激光与光电子学进展,2017,54(8):081002.
[4] Huang H,He K,Zheng X L,et al.Spatial-spectralfeature extraction of hyperspectral image based ondeep learning[J].Laser &Optoelectronics Progress,2017,54(10):101001.黄鸿,何凯,郑新磊,等.基于深度学习的高光谱图像空-谱联合特征提取[J].激光与光电子学进展,2017,54(10):101001.
[5] Hou B H,Yao M L,Wang R,et al.Spatial-spectralsemi-supervised local discriminant analysis forhyperspectral image classification[J].Acta OpticaSinica,2017,37(7):0728002.侯榜焕,姚敏立,王榕,等.面向高光谱图像分类的空谱半监督局部判别分析[J].光学学报,2017,37(7):0728002.
[6] Li D R,Zhang L P,Xia G S.Automatic analysisandmining of remote sensing big data[J].ActaGeodaetica et Cartographica Sinica,2014,43(12):1211-1216.李德仁,张良培,夏桂松.遥感大数据自动分析与数据挖掘[J].测绘学报,2014,43(12):1211-1216.
[7] Lu G F,Wang Y,Zou J,et al.Matrix exponentialbased discriminant locality preserving projections forfeature extraction[J].Neural Networks,2018,97:127-136.
[8] Sellami A,Farah I R.A spatial hypergraph basedsemi-supervised band selection method forhyperspectral imagery semantic interpretation[J].World Academy of Science, Engineering andTechnology,International Journal of Computer,Electrical, Automation, Control and InformationEngineering,2016,10(10):1839-1846.
[9] Zhang B. Advancement of hyperspectral imageprocessing and information extraction[J].Journal ofRemote Sensing,2016,20(5):1062-1090.张兵.高光谱图像处理与信息提取前沿[J].遥感学报,2016,20(5):1062-1090.
[10] Gan Y H,Luo F L,Liu J H,et al.Featureextraction based multi-structure manifold embeddingfor hyperspectral remote sensing image classification[J].IEEE Access,2017,5:25069-25080.
[11] Lodha S P,Kamlapur S.Dimensionality reductiontechniques for hyperspectral images[J].InternationalJournal of Application or Innovation in Engineering &Management(IJAIEM),2014,3(10):92-99.
[12] Zhang X R, Huyan N,Zhou N,et al.Semi-supervised sparse dimensionality reduction forhyperspectral image classification[C]∥2016IEEERegion 10 Conference, November 22-25,2016,Singapore.New York:IEEE,2016:2830-2833.
[13] Xue S Z,Tan R,Chen X H.Method of kernel-basedsemi-supervised locality preserving projection[J].Journal of Computer Applications,2012,32(8):2235-2237,2244.薛寺中,谈锐,陈秀宏.基于核的半监督的局部保留投影降维方法[J].计算机应用,2012,32(8):2235-2237,2244.
[14] Wold S,Esbensen K,Geladi P.Principal componentanalysis[J].Chemometrics and Intelligent LaboratorySystems,1987,2(1/2/3):37-52.
[15] He X F.Locality preserving projectionsp[D].Chicago:University of Chicago,2005.
[16] Zhang J B, Wang J K,Cai X.Sparse localitypreserving discriminative projections for facerecognition[J].Neurocomputing,2017,260:321-330.
[17] Wan M H,Lai Z H.Multi-manifold locality graphembedding based on the maximum margin criterion(MLGE/MMC)for face recognition[J].IEEEAccess,2017,5:9823-9830.
[18] Huang Y, Tang K,Sun Z.Locality sensitivediscriminant analysis for classification ofhyperspectral data[C]∥2014 7th InternationalCongress on Image and Signal Processing,October14-16,2014,Dalian,China.New York:IEEE,2014:292-296.
[19] He X F,Cai D,Yan S C,et al.Neighborhoodpreserving embedding[C]∥Tenth IEEE InternationalConference on Computer Vision(ICCV’05),October17-21,2005,Beijing,China.New York:IEEE,2005,2:1208-1213.
[20] Wu Z Y,Hou B Y,Wang H B,et al.Semi-supervised sparsity neighboring preserving projection[J].Systems Engineering and Electronics,2018,40(4):934-940.吴振宇,侯冰洋,王辉兵,等.半监督稀疏近邻保持投影[J].系统工程与电子技术,2018,40(4):934-940.
[21] Lu X H,He Z Y,Yi S Y,eta l.Joint of locality-andglobality-preserving projections[J].Signal,Imageand Video Processing,2018,12(3):565-572.
[22] Pan Y S,Wang P F,Huang H,eta l.Hyperspectralremote sensing image classification based on SSLPP[J].Computer Science,2013(S2):333-336,373.潘银松,王攀峰,黄鸿,等.基于SSLPP算法对高光谱遥感影像分类[J].计算机科学,2013(S2):333-336,373.
[23] Du W B,LüM,Hou Q L,et al.Semisuperviseddimension reduction based on pairwise constraintpropagation for hyperspectral images[J].IEEEGeoscience and Remote Sensing Letters,2016,13(12):1880-1884.
[24] Huang H, Huang Y B.Improved discriminantsparsity neighborhood preserving embedding forhyperspectral image classification[J].Neurocomputing,2014,136:224-234.
[25] He F,Wang R,Yu Q,et al.Feature extraction ofhyperspectral images of weighted spatial and spectrallocality preserving projection(WSSLPP)[J].Opticsand Precision Engineering,2017,25(1):263-273.何芳,王榕,于强,等.加权空谱局部保持投影的高光谱图像特征提取[J].光学精密工程,2017,25(1):263-273.