基于超图和卷积神经网络的高光谱图像分类
|
摘要
针对高光谱图像数据维数多,光谱信息和空间信息难以提取的问题,提出了一种基于超图和卷积神经网络的分类算法,依据高光谱图像中像素之间的光谱关系和空间关系构建超图;通过超图构建具有谱空联合特征的样本,将其送入卷积神经网络进行特征提取,实现分类。在3种常用的高光谱数据集上进行实验,于Indian Pines数据集上取得了96.63%的总体分类精度。相比于其他算法,所提算法的分类精度高、速度快,而且避免了传统方法在特征提取和融合时出现的不稳定性,验证了其提取的谱空联合信息对高光谱图像具有更强的特征表达能力。
To solve the problem that hyperspectral image data has many dimensions and it is difficult to extract spectral information and spatial information,a classification algorithm is proposed based on a hypergraph and a convolutional neural network.In this algorithm,the hypergraph is first constructed based on the spectral and spatial relationships among pixels in a hyperspectral image,and then a sample with spectral space joint features is constructed through this hypergraph,which is finally sent to the convolutional neural network for feature extraction and thus the classification is finally achieved.The experiment is performed on three most commonly used hyperspectral datasets and an overall classification accuracy of 96.63% on the Indian Pines dataset is achieved.Compared with other algorithms,the proposed algorithm has a high classification accuracy and a high speed,which avoids the instability in feature extraction and fusion by traditional methods.It is verified that the spectral space joint information extracted by the proposed algorithm has a strong feature expression of hyperspectral images.
To solve the problem that hyperspectral image data has many dimensions and it is difficult to extract spectral information and spatial information,a classification algorithm is proposed based on a hypergraph and a convolutional neural network.In this algorithm,the hypergraph is first constructed based on the spectral and spatial relationships among pixels in a hyperspectral image,and then a sample with spectral space joint features is constructed through this hypergraph,which is finally sent to the convolutional neural network for feature extraction and thus the classification is finally achieved.The experiment is performed on three most commonly used hyperspectral datasets and an overall classification accuracy of 96.63% on the Indian Pines dataset is achieved.Compared with other algorithms,the proposed algorithm has a high classification accuracy and a high speed,which avoids the instability in feature extraction and fusion by traditional methods.It is verified that the spectral space joint information extracted by the proposed algorithm has a strong feature expression of hyperspectral images.
引文
[1]Zha Z J,Hua X S,Mei T,et al.Joint multi-label multi-instance learning for image classification[C]∥2008 IEEE Conference on Computer Vision and Pattern Recognition,June 23-28,2008,Anchorage,AK,USA.New York:IEEE,2008:4587384.
[2]Wang J,Chang C I.Independent component analysisbased dimensionality reduction with applications in hyperspectral image analysis[J].IEEE Transactions on Geoscience and Remote Sensing,2006,44(6):1586-1600.
[3]Marconcini M,Camps-Valls G,Bruzzone L.Acomposite semisupervised SVM for classification of hyperspectral images[J].IEEE Geoscience and Remote Sensing Letters,2009,6(2):234-238.
[4]Archibald R,Fann G.Feature selection and classification of hyperspectral images with support vector machines[J].IEEE Geoscience and Remote Sensing Letters,2007,4(4):674-677.
[5]Ma L,Crawford M M,Tian J W.Local manifold learning-based k-nearest-neighbor for hyperspectral image classification[J].IEEE Transactions on Geoscience and Remote Sensing,2010:4099-4109.
[6]Camps-Valls G,Bruzzone L.Kernel-based methods for hyperspectral image classification[J].IEEETransactions on Geoscience and Remote Sensing,2005,43(6):1351-1362.
[7]Berge A,Schistad Solberg A H.Structured Gaussian components for hyperspectral image classification[J].IEEE Transactions on Geoscience and Remote Sensing,2006,44(11):3386-3396.
[8]Huang Y C,Liu Q S,Zhang S T,et al.Image retrieval via probabilistic hypergraph ranking[C]∥2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition,June 13-18,2010,San Francisco,CA,USA.New York:IEEE,2010:5540012.
[9]Bu J J,Tan S L,Chen C,et al.Music recommendation by unified hypergraph[C]∥Proceedings of the international conference on Multimedia-MM′10,October 25-29,2010,Firenze,Italy,2010.New York:ACM,2010:391-400.
[10]Sohn Y,Rebello N S.Supervised and unsupervised spectral angle classifiers[J].Photogrammetric Engineering&Remote Sensing,2002,68(22):1271-1280.
[11]Chang C C,Lin C J.LIBSVM:a library for support vector machines[J].ACM Transactions on Intelligent Systems and Technology,2011,2(3):27.
[12]Hu W,Huang Y Y,Wei L,et al.Deep convolutional neural networks for hyperspectral image classification[J].Journal of Sensors,2015,2015:258619.
[13]Lee H,Kwon H.Going deeper with contextual CNNfor hyperspectral image classification[J].IEEETransactions on Image Processing,2017,26(10):4843-4855.
[14]Ran L Y,Zhang Y N,Wei W,et al.Ahyperspectral image classification framework with spatial pixel pair features[J].Sensors,2017,17(10):2421.
[2]Wang J,Chang C I.Independent component analysisbased dimensionality reduction with applications in hyperspectral image analysis[J].IEEE Transactions on Geoscience and Remote Sensing,2006,44(6):1586-1600.
[3]Marconcini M,Camps-Valls G,Bruzzone L.Acomposite semisupervised SVM for classification of hyperspectral images[J].IEEE Geoscience and Remote Sensing Letters,2009,6(2):234-238.
[4]Archibald R,Fann G.Feature selection and classification of hyperspectral images with support vector machines[J].IEEE Geoscience and Remote Sensing Letters,2007,4(4):674-677.
[5]Ma L,Crawford M M,Tian J W.Local manifold learning-based k-nearest-neighbor for hyperspectral image classification[J].IEEE Transactions on Geoscience and Remote Sensing,2010:4099-4109.
[6]Camps-Valls G,Bruzzone L.Kernel-based methods for hyperspectral image classification[J].IEEETransactions on Geoscience and Remote Sensing,2005,43(6):1351-1362.
[7]Berge A,Schistad Solberg A H.Structured Gaussian components for hyperspectral image classification[J].IEEE Transactions on Geoscience and Remote Sensing,2006,44(11):3386-3396.
[8]Huang Y C,Liu Q S,Zhang S T,et al.Image retrieval via probabilistic hypergraph ranking[C]∥2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition,June 13-18,2010,San Francisco,CA,USA.New York:IEEE,2010:5540012.
[9]Bu J J,Tan S L,Chen C,et al.Music recommendation by unified hypergraph[C]∥Proceedings of the international conference on Multimedia-MM′10,October 25-29,2010,Firenze,Italy,2010.New York:ACM,2010:391-400.
[10]Sohn Y,Rebello N S.Supervised and unsupervised spectral angle classifiers[J].Photogrammetric Engineering&Remote Sensing,2002,68(22):1271-1280.
[11]Chang C C,Lin C J.LIBSVM:a library for support vector machines[J].ACM Transactions on Intelligent Systems and Technology,2011,2(3):27.
[12]Hu W,Huang Y Y,Wei L,et al.Deep convolutional neural networks for hyperspectral image classification[J].Journal of Sensors,2015,2015:258619.
[13]Lee H,Kwon H.Going deeper with contextual CNNfor hyperspectral image classification[J].IEEETransactions on Image Processing,2017,26(10):4843-4855.
[14]Ran L Y,Zhang Y N,Wei W,et al.Ahyperspectral image classification framework with spatial pixel pair features[J].Sensors,2017,17(10):2421.