Probabilistic Methods for Hyperspectral Image

Day - Time: 06 February 2017, h.14:30
Place: Area della Ricerca CNR di Pisa - Room: I-07a
  • Koray Kayabol (Gebze Institute of Technology, Turchia)

Ercan Engin Kuruoglu


As one of the main topics of remote sensing research, hyperspectral images (HSIs) are used in many real-life applications including forest vegetation mapping and classification, urban and land usage, determination of the water resources, and classification of the crop species. In this seminar, we share the results of our research on spectral and spatial classification of hyperspectral images. We use generative and discriminative probabilistic models and Bayesian methods in our research. Three main topics will be presented in the seminar: 1) Spectral-spatial classification using Gaussian mixture models, 2) Low dimensional representation and classification using mixture of probabilistic PCA models and 3) Sparsity constrained classification using multinomial logistic regression.