SVM and MRF-Based Method for Accurate Classification of Hyperspectral Images

Yuliya Tarabalka (1, 2, 3) , Mathieu Fauvel (4) , Jocelyn Chanussot (5) , Jon Atli Benediktsson (2)

1 GSFC - NASA Goddard Space Flight Center
2 University of Iceland [Reykjavik]
3 GIPSA-lab - Grenoble Images Parole Signal Automatique
4 MISTIS - Modelling and Inference of Complex and Structured Stochastic Systems
5 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique

The high number of spectral bands acquired by hyperspectral sensors increases the capability to distinguish physical materials and objects, presenting new challenges to image analysis and classification. This letter presents a novel method for accurate spectral-spatial classification of hyperspectral images. The proposed technique consists of two steps. In the first step, a probabilistic support vector machine pixelwise classification of the hyperspectral image is applied. In the second step, spatial contextual information is used for refining the classification results obtained in the first step. This is achieved by means of a Markov random field regularization. Experimental results are presented for three hyperspectral airborne images and compared with those obtained by recently proposed advanced spectral-spatial classification techniques. The proposed method improves classification accuracies when compared to other classification approaches.

Mots clés: support vector machine (SVM); Markov random field (MRF); hyperspectral images; Classification

Domaines: Traitement des images [eess.IV]

Yuliya Tarabalka, Mathieu Fauvel, Jocelyn Chanussot, Jon Atli Benediktsson. SVM and MRF-Based Method for Accurate Classification of Hyperspectral Images. IEEE Geoscience and Remote Sensing Letters, 2010, 7 (4), pp.736-740. ⟨10.1109/LGRS.2010.2047711⟩. ⟨hal-00578864⟩