Abstract
This paper addresses the automatic inference of a Gibbs distribution dedicated to segment grouping through relaxation labeling. The behavior of this method is studied through the detection of a road-like network from a noisy set of segments extracted from an image during a preprocessing step. Linking segments are added to this set to recover lost road parts. The whole segment set is organized in a relational graph and the road network restoration is modeled as a labeling process. The solution is defined as the labeling maximizing a Gibbs distribution constructed from a set of local costs computed for each graph clique. These cost functions, corresponding to interaction potentials, are learned automatically using multi-layer perceptrons. Supervised learning is performed over a training data set using only binary teaching output, “good” or “bad” configuration example. Several neural networks are used to overcome the problem of the variable complexity of clique configurations.
Chapter PDF
Keywords
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.
References
S. Geman and D. Geman. Stochastic relaxation, Gibbs distribution and the Bayesian restoration of images. IEEE PAMI, 6:721–741, 1984.
I.Y. Kim and H.S. Yang. Efficient image labeling based on Markov random field and error backpropagation network. Pattern Recognition, 26(11):1695–1707, 1993.
J. Kittler and J. Illingworth. Relaxation labeling algorithms-a review. Image Vision Comput., 3(4):206–216, 1985.
A. Manceaux-Demiau, J.-F. Mangin, J. Régis, O. Pizzato, and V. Frouin. Differential features of cortical folds. CVRMed-MRCAS'97 LNCS 1205, Springer Verlag, pages 439–448, 1997.
J.-F. Mangin, J. Régis., I. Bloch, V. Frouin, Y. Samson, and J. Lopez-Krahe. A MRF based random graph medelling the human cortical topography. CVRMed'95, LNCS 905, Springer, pages 177–183, 1995.
W. S. McCulloch and W. A. Pitts. A logical calculus of the ideas immanent in nervous activity. Bulletin of Mathematical Biophysics, 5:115–133, 1943.
F. Rosenblatt. The perceptron: a probabilistic model for information storage and organization in the brain. Psychological Review, 65:386–408, 1958.
F. Tupin, H. Maitre, J.-F. Mangin, J.-M. Nicolas, and E. Pechersky. Linear feature detection on SAR images: Application to the road network. IEEE Geoscience and Remote Sensing, in press, 1997.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1998 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Riviére, D., Mangin, J.F., Martinez, J.M., Chavand, F., Frouin, V. (1998). Neural network based learning of local compatibilities for segment grouping. In: Amin, A., Dori, D., Pudil, P., Freeman, H. (eds) Advances in Pattern Recognition. SSPR /SPR 1998. Lecture Notes in Computer Science, vol 1451. Springer, Berlin, Heidelberg. https://doi.org/10.1007/BFb0033253
Download citation
DOI: https://doi.org/10.1007/BFb0033253
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-64858-1
Online ISBN: 978-3-540-68526-5
eBook Packages: Springer Book Archive