In: Wagner W., Szekely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
153
5. CONCLUSION
In this paper, we applied the model proposed in (Feitosa et al.,
2009) on a set of very high resolutionIKONOS II images of
urban areas within the city of Rio de Janeiro, Brazil.
The results are consistent with the ones presented in (Feitosa et
al., 2009), where series of Landsat images over an agricultural
area were subjected to multitemporal interpretation.
The experiments results presented here and in (Feitosa et al.,
2009) indicate that the multitemporal classification design
based on fuzzy Markov chains generally brings an accuracy
gain in relation to the monotemporal approach. Furthermore, it
has been shown that the more accurate the information coming
from the earlier date, the higher is its contribution to the
multitemporal classification performance. In fact, the fuzzy
Markov chain method seem to be particularly beneficial
whenever there is information regarding the earlier date (t) that
is significantly more accurate than the available information
about the later date (t+1).
We should recall that the assumption underlying the proposed
multitemporal classification model is the existence of a
significant temporal correlation between the data sets. If an
application does not meet this condition, the method is not
expected to work properly at all.
Future research should tackle a number of other important
issues. Experiments should be performed with more time points.
In this case, conditions on the length of the Markov chain could
also be investigated. Broader series of experiments with other
geographic regions and types of images should also be carried
out.
It would be also interesting to investigate other optimization
algorithms, since genetic algorithms spend too much processing
time and do not guarantee that the global optimum solution is
found. A possible candidate for this optimization task could be
some sort of least squares-based algorithm.
PIMAR, 2010. Projeto PIMAR (Programa Integrado de
Monitoria Remota de Fragmentos Florestais e de Crescimento
Urbano no Rio de Janeiro).
http://www.nima.puc-rio.br/sobre_nima/projetos/pimar
(accessed 15 May 2010)
Schmiedle, F., Drechsler, N., Grosse, D., Drechsler, R., 2002.
Heuristic learning based on genetic programming. Genetic
Programming and Evolvable Machines 3 (4), 363-388.
Swain, P. H. 1978. Bayesian classification in a time-varying
environment. IEEE Trans. Sys. Man. and Cybern. 12, 879-883.
Weismiller, R. A., Kristoof, S. J., Scholz, D. K., Anuta, P. E.,
Momen, S. A., 1977. Change detection in coastal zone
environments. Photogrammetric Engineering and Remote
Sensing 43, 1533-1539.
ACKNOWLEDGEMENTS
This work was supported by CNPq (Brazilian National Counsel
of Technological and Scientific Development), FINEP
(Brazilian Innovation Agency) and FAPERJ (Carlos Chagas
Filho Research Support Agency of the State of Rio de Janeiro).
We would also like to thank the Government of the State of Rio
de Janeiro, Secretaria de Estado do Ambiente (SEA) for funding
the PIMAR project.
REFERENCES
Avranchenkov, K. E., Sanchez, E., 2002. Fuzzy Markov chains
and decision-making. Fuzzy Optimization and Decision Making
1 (2), 143-159.
Baatz, M., Schäpe, A. 2000. Multiresolution segmentation - an
optimization approach for high quality multi-scale image
segmentation. In: Strobl, J., Blaschke, T. Angewandte
Geographische Informationsverarbeitung XII. Beiträge zum
AGIT Symposium Salzburg 2000, Herbert Wichmann Verlag,
Karlsruhe.
Feitosa, R. Q., Costa, G. A. O. P., Mota, G. L. A., Feijö, B.,
2009. Cascade classification of Multitemporal. ISPRS
International Journal on Photogrammetry and Remote Sensing
64 (2), 159-170.
Mota, G. L. A., Feitosa, R. Q., Coutinho, H. L. C., Liedtke, C.
E., Müller, S., Pakzad, K., Meirelles, M. S. P., 2007.
Multitemporal fuzzy classification model based on class
transition possibilities. ISPRS International Journal of
Photo grammetry and Remote Sensing 62, 186-200.
