CIEM 05476
CENTRO DE INVESTIGACION Y ESTUDIOS DE MATEMATICA
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Markov Models for image classification.
Autor/es:
FLESIA, A. G.; BUSTOS.O.H.
Lugar:
Maresias, San Pablo Brazil
Reunión:
Otro; XIII Brazilian School of Probability.; 2009
Institución organizadora:
Instituto de Matematica e Estatistica. USP. Brazil
Resumen:
We address the problem of classifying objects in multi-spectral opticalimagery. We discuss two widely used methods, the pointwiseGaussian Maximum Likelihood (ML) and contextual ICM(Iterated Conditional Modes). We show on real data thatthe ICM method is superior, even using the isotropic Pottsmodel for contextual characterization. We discuss generalizationsto the anisotropic model.
