László Czúni (University of Veszprém, Dep. of Image Proc. & Neurocomp., Veszprém, Hungary)

Tamás Szirányi (Computer and Automation Institute, Hungarian Academy of Sciences, Hungary)

Josiane Zerubia (INRIA, Sophia-Antipolis, France, zerubia@sophia.inria.fr)

Abstract:

Due to the large computation power needed for Markovian Random Field (MRF) based image processing, new variations of basic MRF models are implemented. The Cellular Neural Network (CNN) architecture, implemented in real VLSI circuits, is of superior speed in image processing. This very fast CNN can implement the ideas of existing MRF models, which would result in real time processing of images. This VLSI solution gives new tasks since the CNN has a special local architecture. A type of MRF image segmentation with Modified Metropolis Dynamics (MMD) can be well implemented in the CNN architecture. In this paper, we address the improvement of the existing CNN method. We have tried out different multigrid models and compared segmentation results. The main reason for this research is to find the proper implementation of the CNN-MRF technique on CNNs according to the abilities of today's and future's VLSI CNN systems.

Category: segmentation and grouping

What is the original contribution of this work? What is the most closely related work by others and how does this work differ?

This work investigates some multigrid image segmentation methods with the help of the CNNs. We review the implementation of some basic multigrid models. We show test results illustrating segmentation performance and necessary hardware requirements which differ from the previously implemented models and can be a cardinal point in future's CNN chips. These multigrid techniques may trigger the monogrid model in certain circumstances.

• Z. Kató, J. Zerubia, M. Berthod: Satellite image classification using a modified Metropolis dynamics, Proc. ICASSP, San Francisco, March 1992
• Z. Kató, J. Zerubia, M. Berthod: A Hierarchical Markov Random Field Model and Multitemperature Annealing for Parallel Classification, Graphical Models and Image Processing, Vol. 58, No. 1, pp. 18-37, January, 1996
• T. Szirányi, J. Zerubia, D. Geldreich, Z. Kato: Cellular Neural Network in Markov Random Field Image Segmentation , CNNA'96, IEEE, Seville, 1996
• T. Szirányi, J. Zerubia: Markov Random Field Image Segmentation using Cellular Neural Network, IEEE Tr. Circuits and Systems I., Vol. 44, January 1997
• Szirányi, L. Czúni: 'Picture Segmentation with Introducing an Anisotropic Preliminary Step to an MRF Model with Cellular Neural Networks', Proc. of the 13^th ICPR, IEEE, Vienna, 1996