Spectral-Spatial Classification of Hyperspectral Data based on a Stochastic Minimum Spanning Forest Approach

Kevin Bernard (1) , Yuliya Tarabalka (2, 3) , Jesus Angulo (4) , Jocelyn Chanussot (5) , Jon Atli Benediktsson (1)

1 University of Iceland [Reykjavik]
2 GSFC - NASA Goddard Space Flight Center
3 AYIN - Models of spatio-temporal structure for high-resolution image processing
4 CMM - Centre de Morphologie Mathématique
5 GIPSA-SIGMAPHY - GIPSA - Signal Images Physique

In this paper, a new method for supervised hyperspectral data classification is proposed. In particular, the notion of stochastic Minimum Spanning Forest (MSF) is introduced. For a given hyperspectral image, a pixelwise classification is first performed. From this classification map, M marker maps are generated by randomly selecting pixels and labeling them as markers for the construction of Minimum Spanning Forests. The next step consists in building an MSF from each of the M marker maps. Finally, all the M realizations are aggregated with a maximum vote decision rule, in order to build the final classification map. The proposed method is tested on three different data sets of hyperspectral airborne images with different resolutions and contexts. The influence of the number of markers and of the number of realizations M on the results are investigated in experiments. The performance of the proposed method is compared to several classification techniques (both pixelwise and spectral-spatial) using standard quantitative criteria and visual qualitative evaluation.

Domaines: Traitement des images [eess.IV]

Kevin Bernard, Yuliya Tarabalka, Jesus Angulo, Jocelyn Chanussot, Jon Atli Benediktsson. Spectral-Spatial Classification of Hyperspectral Data based on a Stochastic Minimum Spanning Forest Approach. IEEE Transactions on Image Processing, 2012, 21 (4), pp.2008-2021. ⟨10.1109/TIP.2011.2175741⟩. ⟨hal-00728498⟩