Abstract
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.
Article PDF
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.Avoid common mistakes on your manuscript.
References
Yuan Yuan, Haobo Lv, and Xiaoqiang Lu, “Semisupervised change detection method for multi-temporal hyperspectral images,” Neurocomputing 148, 363–375 (2015).
Junhwa Chi and M. M. Crawford, “Selection of landmark points on nonlinear manifolds for spectral unmixing using local homogeneity,” IEEE Trans. Geosci. Remote Sensing Lett. 10 (4), 711–715 (2013).
Lefei Zhang, Liangpei Zhang, Dacheng Tao, and Xin Huang, “On combining multiple features for hyperspectral remote sensing image classification,” IEEE Trans. Geosci. Remote Sensing 50 (3), 879–893 (2012).
Hongjun Su, Yehua Sheng, Peijun Du, Chen Chen, and Kui Liu, “Hyperspectral image classification based on volumetric texture and dimensionality reduction,” Frontiers Earth Sci. 9 (2), 225–236 (2015).
Qiyue Yin, Shu Wu, Ran He, and Liang Wang, “Multiview clustering via pairwise sparse subspace representation,” Neurocomputing 156, 12–21 (2015).
M. Volpi, G. Matasci, M. Kanevski, and D. Tuia, “Semi-supervised multiview embedding for hyperspectral data classification,” Neurocomputing 145, 427–437 (2014).
M. Gnen, “Coupled dimensionality reduction and classification for supervised and semi-supervised multilabel learning,” Pattern Recogn. Lett. 38, 132–141 (2014).
Jun Yu, Dacheng Tao, Yong Rui, and Jun Cheng, “Pairwise constraints based multiview features fusion for scene classification,” Pattern Recogn. 46 (2), 483–496 (2013).
Shuhan Chen, Weiren Shi, and Xiao Lv, “Feature coding for image classification combining global saliency and local difference,” Pattern Recogn. Lett. 51, 44–49 (2015).
S. Hemissi, I. R. Farah, K. Saheb Ettabaa, and B. Solaiman, “Multi-spectro-temporal analysis of hyperspectral imagery based on 3D spectral modeling and multilinear algebra,” IEEE Trans. Geosci. Remote Sensing 51 (1), 199–216 (2013).
Bor-Chen Kuo and Cheng-Hsuan Li, “Kernel nonparametric weighted feature extraction for classification,” in AI 2005: Advances in Artificial Intelligence, Ed. by Shichao Zhang and Ray Jarvis (Springer, Berlin, Heidelberg, 2005), pp. 567–576.
M. Dalla Mura, A. Villa, J. A. Benediktsson, J. Chanussot, and L. Bruzzone, “Classification of hyperspectral images by using extended morphological attribute profiles and independent component analysis,” IEEE Trans. Geosci. Remote Sensing Lett. 8 (3), 542–546 (2011).
L. Journaux, M.-F. Destain, J. Miteran, A. Piron, and F. Cointault, “Texture classification with generalized Fourier descriptors in dimensionality reduction context: an overview exploration,” in Artificial Neural Networks in Pattern Recognition, Ed. by L. Prevost, S. Marinai, and F. Schwenker (Springer, Berlin, Heidelberg, 2008), pp. 280–291.
L. Yan and D. P. Roy, “Improved time series land cover classification by missing-observation-adaptive nonlinear dimensionality reduction,” Remote Sensing Environ. 158, 478–491 (2015).
Jun Li, J. M. Bioucas-Dias, and A. Plaza, “Semisupervised hyperspectral image classification using soft sparse multinomial logistic regression,” IEEE Trans. Geosci. Remote Sensing Lett. 10 (2), 318–322 (2013).
A. Nealen, T. Igarashi, O. Sorkine, and M. Alexa, “Laplacian mesh optimization,” in Proc. 4th ACM Int. Conf. on Computer Graphics and Interactive Techniques in Australasia and Southeast Asia, GRAPHITE’06 (New York, 2006), pp. 381–389.
Takashi Kuremoto, Shinsuke Kimura, Kunikazu Kobayashi, and Masanao Obayashi, “Time series forecasting using a deep belief network with restricted Boltzmann machines,” Neurocomputing 137, 47–56 (2014).
H. Larochelle, M. Mandel, R. Pascanu, and Y. Bengio, “Learning algorithms for the classification restricted Boltzmann machine,” J. Mach. Learn. Res. 13 (1), 643–669 (2012).
J. Louradour and H. Larochelle, “Classification of sets using restricted Boltzmann machines,” CoRR, abs/1103.4896, 2011.
Liu Jian-wei, Chi Guang-hui, and Luo Xiong-lin, “Contrastive divergence learning of restricted Boltzmann machine,” in Proc. 2nd IEEE Computer Soc. Int. Conf. on Electric Technology and Civil Engineering, ICETCE’12 (Washington, 2012), pp. 712–715.
Author information
Authors and Affiliations
Corresponding author
Additional information
This paper uses the materials of the report submitted at the 9th Open German-Russian Workshop on Pattern Recognition and Image Understanding, held in Koblenz, December 1–5, 2014 (OGRW-9-2014).
The article is published in the original.
Selim Hemissi received the PHD degree in computer sciences and signal processing jointly from Telecom Bretagne, Brest, France and the Ecole Nationale des Sciences de Informatique (ENSI), Manouba, Tunisia, in 2014. He is a Permanent Researcher at Laboratory RIADI, University of Manouba, since 2008. His work is mainly related with pattern recognition, signal processing, and machine learning applied to remote sensing hyperspectral images. He also enjoyed a research partnership with the Department ITI in Telecom Bretagne where he is currently an Associate Researcher. Mr. Hemissi is a member of Arts-Pi Tunisia and IEEE Student Brunch of Telecom Bretagne.
Imed Riadh Farah received the MD degree from the ISG Institute of Computer Sciences, Tunis, Tunisia, in 1995, and the Dr. Eng. degree from the Ecole Nationale des Sciences de l’Informatique (ENSI), Manouba, Tunisia, in 2003. After working as a Research Assistant (from 1996) and a Permanent Researcher at Laboratory RIADI, EXSI National School of Computer Sciences engineering (since 1995), he has been an Associate Professor at the University of Manouba, since 2010. He has been an Associate Researcher in the Department ITI-Telecom Bretagne, Brest, France, since January 2009. His research interests include image processing, pattern recognition, artificial intelligence, data mining, and their application to remote sensing. Currently, he is the Director of the Higher Institute of Arts and Multimedia. Dr. Farah is a member of Arts-Pi Tunisia.
Rights and permissions
About this article
Cite this article
Hemissi, S., Farah, I.R. Efficient multi-temporal hyperspectral signatures classification using a Gaussian-Bernoulli RBM based approach. Pattern Recognit. Image Anal. 26, 190–196 (2016). https://doi.org/10.1134/S1054661816010211
Received:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S1054661816010211