
Publications of D. Geldreich
Result of the query in the list of publications :
Article 
1  Image segmentation using Markov random field model in fully parallel cellular network architectures. T. Szirányi and J. Zerubia and L. Czúni and D. Geldreich and Z. Kato. Real Time Imaging, 6(3): pages 195211, June 2000.
@ARTICLE{jz00y,

author 
= 
{Szirányi, T. and Zerubia, J. and Czúni, L. and Geldreich, D. and Kato, Z.}, 
title 
= 
{Image segmentation using Markov random field model in fully parallel cellular network architectures}, 
year 
= 
{2000}, 
month 
= 
{June}, 
journal 
= 
{Real Time Imaging}, 
volume 
= 
{6}, 
number 
= 
{3}, 
pages 
= 
{195211}, 
pdf 
= 
{http://dx.doi.org/10.1006/rtim.1998.0159}, 
keyword 
= 
{} 
} 
Abstract :
Markovian approaches to early vision processes need a huge amount of computing power. These algorithms can usually be implemented on parallel computing structures. Herein, we show that the Markovian labeling approach can be implemented in fully parallel cellular network architectures, using simple functions and data representations. This makes possible to implement our model in parallel imaging VLSI chips.
As an example, we have developed a simplified statistical image segmentation algorithm for the Cellular Neural/Nonlinear Networks Universal Machine (CNNUM), which is a new image processing tool, containing thousands of cells with analog dynamics, local memories and processing units. The Modified Metropolis Dynamics (MMD) optimization method can be implemented into the raw analog architecture of the CNNUM. We can introduce the whole pseudostochastic segmentation process in the CNN architecture using 8 memories/cell. We use simple arithmetic functions (addition, multiplication), equalitytest between neighboring pixels and very simple nonlinear output functions (step, jigsaw). With this architecture, the proposed VLSI CNN chip can execute a pseudostochastic relaxation algorithm of about 100 iterations in about 100 μs.
In the suggested solution the segmentation is unsupervised, where a pixellevel statistical estimation model is used. We have tested different monogrid and multigrid architectures.
In our CNNUM model several complex preprocessing steps can be involved, such as textureclassification or anisotropic diffusion. With these preprocessing steps, our fully parallel cellular system may work as a highlevel image segmentation machine, using only simple functions based on the closeneighborhood of a pixel. 

top of the page
These pages were generated by
