Efficient multi-temporal hyperspectral signatures classification using a Gaussian-Bernoulli RBM based approach

This paper presents an efficient Gaussian-Bernoulli Restricted Boltzmann Machines (GB-RBM) framework in order to better address the classification challenge of remotely sensed images. The proposed approach relies on generating well-designed features for a new 3D modality of spectral signature. For this purpose, mesh smoothing is introduced to reduce noise while conserving the main geometric features of the multi-temporal spectral signature. Then, we propose the use of an RBM (Restricted Boltzmann Machine) framework as stand-alone non-linear classifier. The adapted framework focuses on a cooperative integrated generative-discriminative objective allowing the integration of modeling input features and their classification process in one-pass algorithm. The main benefit of the proposed approach is the ability to learn more discriminative features. We evaluated our approach within different scenarios and we demonstrated its usefulness for noisy high dimensional hyperspectral images.

[1]  Johel Mitéran,et al.  Texture Classification with Generalized Fourier Descriptors in Dimensionality Reduction Context: An Overview Exploration , 2008, ANNPR.

[2]  Liangpei Zhang,et al.  On Combining Multiple Features for Hyperspectral Remote Sensing Image Classification , 2012, IEEE Transactions on Geoscience and Remote Sensing.

[3]  Xiaoqiang Lu,et al.  Semi-supervised change detection method for multi-temporal hyperspectral images , 2015, Neurocomputing.

[4]  Imed Riadh Farah,et al.  Multi-Spectro-Temporal Analysis of Hyperspectral Imagery Based on 3-D Spectral Modeling and Multilinear Algebra , 2013, IEEE Transactions on Geoscience and Remote Sensing.

[5]  Peijun Du,et al.  Hyperspectral image classification based on volumetric texture and dimensionality reduction , 2015, Frontiers of Earth Science.

[6]  Hugo Larochelle,et al.  Classification of Sets using Restricted Boltzmann Machines , 2011, UAI.

[7]  Jon Atli Benediktsson,et al.  Classification of Hyperspectral Images by Using Extended Morphological Attribute Profiles and Independent Component Analysis , 2011, IEEE Geoscience and Remote Sensing Letters.

[8]  Marc Alexa,et al.  Laplacian mesh optimization , 2006, GRAPHITE '06.

[9]  Lin Yan,et al.  Improved time series land cover classification by missing-observation-adaptive nonlinear dimensionality reduction , 2015 .

[10]  Liang Wang,et al.  Multi-view clustering via pairwise sparse subspace representation , 2015, Neurocomputing.

[11]  Mehmet Gönen,et al.  Coupled dimensionality reduction and classification for supervised and semi-supervised multilabel learning , 2014, Pattern Recognit. Lett..

[12]  Antonio J. Plaza,et al.  Semisupervised Hyperspectral Image Classification Using Soft Sparse Multinomial Logistic Regression , 2013, IEEE Geoscience and Remote Sensing Letters.

[13]  Melba M. Crawford,et al.  Selection of Landmark Points on Nonlinear Manifolds for Spectral Unmixing Using Local Homogeneity , 2013, IEEE Geoscience and Remote Sensing Letters.

[14]  Liu Jian-wei,et al.  Contrastive Divergence Learning of Restricted Boltzmann Machine , 2012 .

[15]  Jun Yu,et al.  Pairwise constraints based multiview features fusion for scene classification , 2013, Pattern Recognit..

[16]  Weiren Shi,et al.  Feature coding for image classification combining global saliency and local difference , 2015, Pattern Recognit. Lett..

[17]  Razvan Pascanu,et al.  Learning Algorithms for the Classification Restricted Boltzmann Machine , 2012, J. Mach. Learn. Res..

[18]  Michele Volpi,et al.  Semi-supervised multiview embedding for hyperspectral data classification , 2014, Neurocomputing.

[19]  Kunikazu Kobayashi,et al.  Time series forecasting using a deep belief network with restricted Boltzmann machines , 2014, Neurocomputing.

[20]  Bor-Chen Kuo,et al.  Kernel Nonparametric Weighted Feature Extraction for Classification , 2005, Australian Conference on Artificial Intelligence.