基于线性嵌入和张量流形的高光谱特征提取
|
摘要
为了使降维结果更好地体现高光谱数据的空间结构信息,并进一步提高分类精度,提出了一种基于线性嵌入和张量流形的高光谱特征提取算法。不同于其他流形结构的表达方法,所提算法采用协同表示理论求解全局线性嵌入的权重矩阵,更有利于保持高维数据的全局信息,提高了流形结构表达的准确性。同时,建立了基于多特征描述的张量流形降维框架,得到的显式映射具有较强的可靠性和全局适应性。实验结果表明:与主成分分析、局部线性嵌入、拉普拉斯特征映射和线性保留投影等算法相比,所提算法表现出了更优越的分类性能。
In order to express the spatial structure information of hyperspectral image more effectively and improve the classification accuracy after dimensionality reduction, we propose a hyperspectral feature extraction algorithm based on linear embedding and tensor manifold. Different from other manifold structure expression methods, the proposed algorithm uses the cooperative representation theory to solve the weight matrix for globally linear embedding, which is more beneficial to maintain the global information of high dimensional data and improve the accuracy of manifold structure expression. At the same time, the dimension reduction framework of tensor manifold based on multi-feature description is established, and the obtained explicit mapping has strong reliability and global adaptability. Experimental results show that compared with the principal component analysis, locally linear embedding, Laplacian Eigenmap, linearity preserving projection and other algorithms, the proposed algorithm has better classification performance.
In order to express the spatial structure information of hyperspectral image more effectively and improve the classification accuracy after dimensionality reduction, we propose a hyperspectral feature extraction algorithm based on linear embedding and tensor manifold. Different from other manifold structure expression methods, the proposed algorithm uses the cooperative representation theory to solve the weight matrix for globally linear embedding, which is more beneficial to maintain the global information of high dimensional data and improve the accuracy of manifold structure expression. At the same time, the dimension reduction framework of tensor manifold based on multi-feature description is established, and the obtained explicit mapping has strong reliability and global adaptability. Experimental results show that compared with the principal component analysis, locally linear embedding, Laplacian Eigenmap, linearity preserving projection and other algorithms, the proposed algorithm has better classification performance.
引文
[1] Ghamisi P,Yokoya N,Li J,et al.Advances in hyperspectral image and signal processing:a comprehensive overview of the state of the art[J].Proceedings of the IEEE,2017,5(4):37-78.
[2] Bioucas-Dias J M,Plaza A,Camps-Valls G,et al.Hyperspectral remote sensing data analysis and future challenges[J].Proceedings of the IEEE,2013,1(2):6-36.
[3] Li W,Feng F B,Li H C,et al.Discriminant analysis-based dimension reduction for hyperspectral image classification:a survey of the most recent advances and an experimental comparison of different techniques[J].Proceedings of the IEEE,2018,6(1):15-34.
[4] Sun W W,Yang G,Du B,et al.A sparse and low-rank near-isometric linear embedding method for feature extraction in hyperspectral imagery classification[J].Proceedings of the IEEE,2017,55(7):4032-4046.
[5] Dong A G,Li J X,Zhang B,et al.Hyperspectral image classification algorithm based on spectral clustering and sparse representation[J].Acta Optica Sinica,2017,37(8):0828005.董安国,李佳逊,张蓓,等.基于谱聚类和稀疏表示的高光谱图像分类算法[J].光学学报,2017,37(8):0828005.
[6] Zhang W Q,Li X R,Dou Y X,et al.A geometry-based band selection approach for hyperspectral image analysis[J].Proceedings of the IEEE,2018,56(8):4318-4333.
[7] Wang M D,Yu J,Niu L J,et al.Feature extraction for hyperspectral images using low-rank representation with neighborhood preserving regularization[J].Proceedings of the IEEE,2017,14(6):836-840.
[8] Wei F,He M Y,Feng Y,et al.Feature extraction on matrix factorization for hyperspectral data[J].Journal of Infrared and Millimeter Waves,2014,33(6):674-679.魏峰,何明一,冯燕,等.基于矩阵分解的高光谱数据特征提取[J].红外与毫米波学报,2014,33(6):674-679.
[9] Licciardi G,Marpu P R,Chanussot J,et al.Linear versus nonlinear PCA for the classification of hyperspectral data based on the extended morphological profiles[J].Proceedings of the IEEE,2012,9(3):447-451.
[10] Green A A,Berman M,Switzer P,et al.A transformation for ordering multispectral data in terms of image quality with implications for noise removal[J].Proceedings of the IEEE,1988,26(1):65-74.
[11] Crawford M,Held A A,Burger J.Nonlinear manifolds for feature extraction:opportunities and challenge[J].Proceedings of the IEEE,2011:1-3.
[12] Lunga D,Prasad S,Crawford M M,et al.Manifold-learning-based feature extraction for classification of hyperspectral data:a review of advances in manifold learning[J].Proceedings of the IEEE,2014,31(1):55-66.
[13] Ma L,Crawford M M,Yang X Q,et al.Local-manifold-learning-based graph construction for semisupervised hyperspectral image classification[J].Proceedings of the IEEE,2015,53(5):2832-2844.
[14] Zhai Y G,Zhang L F,Wang N,et al.A modified locality-preserving projection approach for hyperspectral image classification[J].Proceedings of the IEEE,2016,13(8):1059-1063.
[15] Chen X Z,Hu H J,Wang X S.Dimensionality reduction for hyperspectral data using weighted neighborhood preserving embedding[J].Journal of China University of Mining & Technology,2013,42(6):1066-1072.陈新忠,胡汇涓,王雪松.基于加权近邻保持嵌入的高光谱数据降维方法[J].中国矿业大学学报,2013,42(6):1066-1072.
[16] He X F,Niyogi P.Locality preserving projection[C].Advances in Neural Information Processing System,2004(16):153-160.
[17] Roweis S T,Saul L K.Nonlinear dimensionality reduction by locally linear embedding[J].Science,2000,290(5500):2323-2326.
[18] Chen Z,Wang B,Zhang L M.Dimensionality reduction and classification based on lower rank tensor analysis for hyperspectral imagery[J].Journal of Infrared and Millimeter Waves,2013,32(6):569-575.陈昭,王斌,张立明.基于低秩张量分析的高光谱图像降维与分类[J].红外与毫米波学报,2013,32(6):569-575.
[19] An J L,Zhang X R,Zhou H Y,et al.Tensor-based low-rank graph with multimanifold regularization for dimensionality reduction of hyperspectral images[J].Proceedings of the IEEE,2018,56(8):4731-4746.
[20] Li W,Du Q.Collaborative representation for hyperspectral anomaly detection[J].Proceedings of the IEEE,2015,53(3):1463-1474.
[21] Sidiropoulos N D,de Lathauwer L,Fu X,et al.Tensor decomposition for signal processing and machine learning[J].Proceedings of the IEEE,2017,65(13):3551-3582.
[22] Deng Y J,Li H C,Fu K,et al.Tensor low-rank discriminant embedding for hyperspectral image dimensionality reduction[J].Proceedings of the IEEE,2018,56(12):7183-7194.
[23] Makantasis K,Doulamis A D,Doulamis N D,et al.Tensor-based classification models for hyperspectral data analysis[J].Proceedings of the IEEE,2018,56(12):6884-6898.
[24] Zhang T H,Tao D C,Li X L,et al.Patch alignment for dimensionality reduction[J].Proceedings of the IEEE,2009,21(9):1299-1313.
[2] Bioucas-Dias J M,Plaza A,Camps-Valls G,et al.Hyperspectral remote sensing data analysis and future challenges[J].Proceedings of the IEEE,2013,1(2):6-36.
[3] Li W,Feng F B,Li H C,et al.Discriminant analysis-based dimension reduction for hyperspectral image classification:a survey of the most recent advances and an experimental comparison of different techniques[J].Proceedings of the IEEE,2018,6(1):15-34.
[4] Sun W W,Yang G,Du B,et al.A sparse and low-rank near-isometric linear embedding method for feature extraction in hyperspectral imagery classification[J].Proceedings of the IEEE,2017,55(7):4032-4046.
[5] Dong A G,Li J X,Zhang B,et al.Hyperspectral image classification algorithm based on spectral clustering and sparse representation[J].Acta Optica Sinica,2017,37(8):0828005.董安国,李佳逊,张蓓,等.基于谱聚类和稀疏表示的高光谱图像分类算法[J].光学学报,2017,37(8):0828005.
[6] Zhang W Q,Li X R,Dou Y X,et al.A geometry-based band selection approach for hyperspectral image analysis[J].Proceedings of the IEEE,2018,56(8):4318-4333.
[7] Wang M D,Yu J,Niu L J,et al.Feature extraction for hyperspectral images using low-rank representation with neighborhood preserving regularization[J].Proceedings of the IEEE,2017,14(6):836-840.
[8] Wei F,He M Y,Feng Y,et al.Feature extraction on matrix factorization for hyperspectral data[J].Journal of Infrared and Millimeter Waves,2014,33(6):674-679.魏峰,何明一,冯燕,等.基于矩阵分解的高光谱数据特征提取[J].红外与毫米波学报,2014,33(6):674-679.
[9] Licciardi G,Marpu P R,Chanussot J,et al.Linear versus nonlinear PCA for the classification of hyperspectral data based on the extended morphological profiles[J].Proceedings of the IEEE,2012,9(3):447-451.
[10] Green A A,Berman M,Switzer P,et al.A transformation for ordering multispectral data in terms of image quality with implications for noise removal[J].Proceedings of the IEEE,1988,26(1):65-74.
[11] Crawford M,Held A A,Burger J.Nonlinear manifolds for feature extraction:opportunities and challenge[J].Proceedings of the IEEE,2011:1-3.
[12] Lunga D,Prasad S,Crawford M M,et al.Manifold-learning-based feature extraction for classification of hyperspectral data:a review of advances in manifold learning[J].Proceedings of the IEEE,2014,31(1):55-66.
[13] Ma L,Crawford M M,Yang X Q,et al.Local-manifold-learning-based graph construction for semisupervised hyperspectral image classification[J].Proceedings of the IEEE,2015,53(5):2832-2844.
[14] Zhai Y G,Zhang L F,Wang N,et al.A modified locality-preserving projection approach for hyperspectral image classification[J].Proceedings of the IEEE,2016,13(8):1059-1063.
[15] Chen X Z,Hu H J,Wang X S.Dimensionality reduction for hyperspectral data using weighted neighborhood preserving embedding[J].Journal of China University of Mining & Technology,2013,42(6):1066-1072.陈新忠,胡汇涓,王雪松.基于加权近邻保持嵌入的高光谱数据降维方法[J].中国矿业大学学报,2013,42(6):1066-1072.
[16] He X F,Niyogi P.Locality preserving projection[C].Advances in Neural Information Processing System,2004(16):153-160.
[17] Roweis S T,Saul L K.Nonlinear dimensionality reduction by locally linear embedding[J].Science,2000,290(5500):2323-2326.
[18] Chen Z,Wang B,Zhang L M.Dimensionality reduction and classification based on lower rank tensor analysis for hyperspectral imagery[J].Journal of Infrared and Millimeter Waves,2013,32(6):569-575.陈昭,王斌,张立明.基于低秩张量分析的高光谱图像降维与分类[J].红外与毫米波学报,2013,32(6):569-575.
[19] An J L,Zhang X R,Zhou H Y,et al.Tensor-based low-rank graph with multimanifold regularization for dimensionality reduction of hyperspectral images[J].Proceedings of the IEEE,2018,56(8):4731-4746.
[20] Li W,Du Q.Collaborative representation for hyperspectral anomaly detection[J].Proceedings of the IEEE,2015,53(3):1463-1474.
[21] Sidiropoulos N D,de Lathauwer L,Fu X,et al.Tensor decomposition for signal processing and machine learning[J].Proceedings of the IEEE,2017,65(13):3551-3582.
[22] Deng Y J,Li H C,Fu K,et al.Tensor low-rank discriminant embedding for hyperspectral image dimensionality reduction[J].Proceedings of the IEEE,2018,56(12):7183-7194.
[23] Makantasis K,Doulamis A D,Doulamis N D,et al.Tensor-based classification models for hyperspectral data analysis[J].Proceedings of the IEEE,2018,56(12):6884-6898.
[24] Zhang T H,Tao D C,Li X L,et al.Patch alignment for dimensionality reduction[J].Proceedings of the IEEE,2009,21(9):1299-1313.