Two Stage SVM Classification for Hyperspectral Data

In this article, we present a method of enhancing the SVM classification of hyperspectral data with the use of three supporting classifiers. It is done by applying the fully trained classifiers on learning set to obtain the pattern of their behavior which then can be used for refinement of classifier construction. The second stage either is a straightforward translation of first stage, if the first stage classifiers agree on the result, or it consists of using retrained SVM classifier with only the data from learning data selected using first stage. The scheme shares some features with committee of experts fusion scheme, yet it clearly distinguishes lead classifier using the supporting ones only to refine its construction. We present the construction of two-stage scheme, then test it against the known Indian Pines HSI dataset and test it against straightforward use of SVM classifier, over which our method achieves noticeable improvement.

[1]  Jonathan Li,et al.  Bayesian Classification of Hyperspectral Imagery Based on Probabilistic Sparse Representation and Markov Random Field , 2014, IEEE Geoscience and Remote Sensing Letters.

[2]  Jon Atli Benediktsson,et al.  Spectral–Spatial Classification of Hyperspectral Images With a Superpixel-Based Discriminative Sparse Model , 2015, IEEE Transactions on Geoscience and Remote Sensing.

[3]  Sunil Bhaskaran,et al.  Integrating imaging spectroscopy (445–2543 nm) and geographic information systems for post-disaster management: a case of hailstorm damage in Sydney , 2004 .

[4]  D. Ruta,et al.  An Overview of Classifier Fusion Methods , 2000 .

[5]  Trac D. Tran,et al.  Hyperspectral Image Classification via Kernel Sparse Representation , 2011, IEEE Transactions on Geoscience and Remote Sensing.

[6]  Ludmila I. Kuncheva,et al.  A Theoretical Study on Six Classifier Fusion Strategies , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[7]  Lorenzo Bruzzone,et al.  Classification of hyperspectral remote sensing images with support vector machines , 2004, IEEE Transactions on Geoscience and Remote Sensing.

[8]  Paolo Gamba,et al.  Comparison of support vector machine-based processing chains for hyperspectral image classification , 2010, Optical Engineering + Applications.

[9]  Da-Wen Sun,et al.  Recent Progress of Hyperspectral Imaging on Quality and Safety Inspection of Fruits and Vegetables: A Review. , 2015, Comprehensive reviews in food science and food safety.

[10]  Antonio J. Plaza,et al.  Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based Approaches , 2012, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing.

[11]  Vidya B. Manian,et al.  Land cover and benthic habitat classification using texture features from hyperspectral and multispectral images , 2007, J. Electronic Imaging.

[12]  David H. Wolpert,et al.  Stacked generalization , 1992, Neural Networks.

[13]  Patrick Haffner,et al.  Support vector machines for histogram-based image classification , 1999, IEEE Trans. Neural Networks.

[14]  C. Brodley,et al.  Decision tree classification of land cover from remotely sensed data , 1997 .

[15]  L. Cooper,et al.  When Networks Disagree: Ensemble Methods for Hybrid Neural Networks , 1992 .

[16]  James C. Bezdek,et al.  Decision templates for multiple classifier fusion: an experimental comparison , 2001, Pattern Recognit..

[17]  Ludovic Denoyer,et al.  Bayesian network model for semi-structured document classification , 2004, Inf. Process. Manag..