by Zoltan Kato, Josiane Zerubia, Mark Berthod
Abstract:
In this paper, we present a pseudo-stochastic variation of the Metropolis dynamics for combinatorial optimization in image classification using Markov Random Fields. At high temperature, the behavior of our algorithm is similar to the stochastic ones. However, if the temperature is less than a certain threshold, it becomes deterministic. The ``length'' of the ``pseudo-stochastic'' phase is controlled by a constant threshold used in the modified dynamics. The algorithm compares favorably to recent stochastic or deterministic methods and yields an approximate but usually good solution to the optimization problem. The algorithm runs on a Connection Machine. It is applied to standard pixel classification problem; objective and subjective comparisons with other algorithms have been made.
Reference:
Zoltan Kato, Josiane Zerubia, Mark Berthod, Satellite Image Classification Using a Modified Metropolis Dynamics, In Proceedings of International Conference on Acoustics, Speech and Signal Processing, volume 3, San-Francisco, California, USA, pp. 573-576, 1992.
Bibtex Entry:
@string{icassp="Proceedings of International Conference on Acoustics, Speech and Signal Processing"}
@InProceedings{Kato-etal92a,
author = {Kato, Zoltan and Zerubia, Josiane and Berthod, Mark},
title = {Satellite Image Classification Using a Modified
{M}etropolis Dynamics},
booktitle = icassp,
pages = {573-576},
year = 1992,
volume = 3,
address = {San-Francisco, California, USA},
month = mar,
organization = {IEEE},
ps = {../papers/icassp92.ps},
abstract = {In this paper, we present a pseudo-stochastic
variation of the Metropolis dynamics for
combinatorial optimization in image classification
using Markov Random Fields. At high temperature, the
behavior of our algorithm is similar to the
stochastic ones. However, if the temperature is less
than a certain threshold, it becomes
deterministic. The ``length'' of the
``pseudo-stochastic'' phase is controlled by a
constant threshold used in the modified
dynamics. The algorithm compares favorably to recent
stochastic or deterministic methods and yields an
approximate but usually good solution to the
optimization problem. The algorithm runs on a
Connection Machine. It is applied to standard pixel
classification problem; objective and subjective
comparisons with other algorithms have been made. }
}