Abstract | ||
---|---|---|
Gray-scale image segmentation can be cast as a combinatorial optimization problem using a Markov Random Field (MRF) model. Local and global search algorithms were proposed to find an optimal solution in a reasonable amount of time. Both local and global search algorithms have their strengths and their shortcomings. The aim of this work is to propose and study a combination between local and global search strategies. For the local search we use the Iterated Conditional Modes (ICM) while the global optimization is performed by an Ant Colony System (ACS). ICM and ACS are both based on a MRF model to assess the local and the global quality of an arbitrary segmentation. The local optimization is performed by combining Maximum Likelihood (ML) for estimation of the parameters and ICM to find the solution given a fixed set of these parameters. A comparative study is performed to select the combination strategy that gives a better result. The effectiveness of our proposal is proven through experimentation. The problem of parameterization of our solution is addressed through an empirical study that shows the effect of each parameter on the behavior of the proposed solution. |
Year | DOI | Venue |
---|---|---|
2021 | 10.1007/s00500-021-05957-1 | SOFT COMPUTING |
Keywords | DocType | Volume |
Markov random fields, Maximum likelihood, Iterated conditional modes, Ant colony system, HyperCube framework | Journal | 25 |
Issue | ISSN | Citations |
15 | 1432-7643 | 0 |
PageRank | References | Authors |
0.34 | 0 | 2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Houcemeddine Filali | 1 | 0 | 0.34 |
Karim Kalti | 2 | 20 | 8.58 |