by Zoltan Kato, Mark Berthod, Josiane Zerubia
Abstract:
In this paper, we are interested in massively parallel multiscale relaxation algorithms applied to image classification. First, we present a classical multiscale model applied to supervised image classification.The model consists of a label pyramid and a whole observation field. The potential functions of the coarse grid are derived by simple computations. Then, we propose another scheme introducing a local interaction between two neighbor grids in the label pyramid. This is a way to incorporate cliques with far apart sites for a reasonable price. Finally we present the results on noisy synthetic data and on a SPOT image obtained by different relaxation methods using these models.
Reference:
Zoltan Kato, Mark Berthod, Josiane Zerubia, Parallel Image Classification using Multiscale Markov Random Fields, In Proceedings of International Conference on Acoustics, Speech and Signal Processing, volume 5, Minneapolis, USA, pp. 137-140, 1993.
Bibtex Entry:
@string{icassp="Proceedings of International Conference on Acoustics, Speech and Signal Processing"}
@InProceedings{Kato-etal93c,
author = {Kato, Zoltan and Berthod, Mark and Zerubia, Josiane},
title = {Parallel Image Classification using Multiscale
{M}arkov Random Fields},
booktitle = icassp,
pages = {137-140},
year = 1993,
volume = 5,
address = {Minneapolis, USA},
month = apr,
organization = {IEEE},
ps = {../papers/icassp93.ps},
abstract = {In this paper, we are interested in massively
parallel multiscale relaxation algorithms applied to
image classification. First, we present a classical
multiscale model applied to supervised image
classification.The model consists of a label pyramid
and a whole observation field. The potential
functions of the coarse grid are derived by simple
computations. Then, we propose another scheme
introducing a local interaction between two neighbor
grids in the label pyramid. This is a way to
incorporate cliques with far apart sites for a
reasonable price. Finally we present the results on
noisy synthetic data and on a SPOT image obtained by
different relaxation methods using these models. }
}