Logo eprints

A Survey of SOM-Based Active Contour Models for Image Segmentation

Abdelsamea, Mohammed and Gnecco, Giorgio and Gaber, Mohamed Medhat A Survey of SOM-Based Active Contour Models for Image Segmentation. In: Advances in Self-Organizing Maps and Learning Vector Quantization. Advances in Intelligent Systems and Computing (295). Springer, pp. 293-302. ISBN 978-3-319-07695-9 (2014)

Full text not available from this repository.


Self Organizing Maps (SOMs) have attracted the attention of many computer vision scientists, particularly when dealing with image segmentation as a contour extraction problem. The idea of utilizing the prototypes (weights) of a SOM to model an evolving contour has produced a new class of Active Contour Models (ACMs), known as SOM-based ACMs. Such models have been proposed in general with the aim of exploiting the specific ability of SOMs to learn the edge-map information via their topology preservation property, and overcoming some drawbacks of other ACMs, such as trapping into local minima of the image energy functional to be minimized in such models. In this survey paper, the main principles of SOMs and their application in modelling active contours are first highlighted. Then, we review existing SOM-based ACMs with a focus on their advantages and disadvantages in modelling the evolving contour via different kinds of SOMs. Finally, some current research directions are identified.

Item Type: Book Section
Identification Number: 10.1007/978-3-319-07695-9_28
Additional Information: Proceedings of the 10th International Workshop, WSOM 2014, Mittweida, Germany, July, 2-4, 2014
Uncontrolled Keywords: Image segmentation; Self Organizing Maps; active contours; SOM-based ACMs; topology preservation; neural networks
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Research Area: Computer Science and Applications
Depositing User: Ms T. Iannizzi
Date Deposited: 18 Feb 2015 14:33
Last Modified: 18 Feb 2015 14:33
URI: http://eprints.imtlucca.it/id/eprint/2610

Actions (login required)

Edit Item Edit Item