Automated Design of a Genetic Algorithm for Image Segmentation Using the Iterated Local Search

Thambo Nyathi

2024

Abstract

Image thresholding is a fundamental technique used in image processing for segmentation. This is the process of determining optimal thresholds for an image. When the number of thresholds exceeds two, that is, multilevel thresholding, the computational complexity of the process increases exponentially. This has resulted in the popularity of addressing this problem by using metaheuristic methods. However, metaheuristics are pa-rameterised and their effectiveness depends on their configuration, which is often performed manually using an iterative trial-and-error approach. This leads to less effective designs that yield less accurate thresholds and longer design times. This study proposes using an Iterated Local Search to configure a low-level metaheuristic, namely, a Genetic Algorithm(GA), to solve the multilevel threshold problem. The performance of the proposed approach was compared with that of a manually designed standard GA approach, and evaluated using T2 weighted Magnetic Resonance images of the brain. Furthermore, the proposed approach is compared with two other metaheuristic algorithms for the same problem. The results showed that the automatically designed genetic algorithm significantly outperformed the standard genetic algorithm approach and the other two algorithms on the set objective function. Although the runtimes were higher than those of the manual design approach, better thresholds were obtained.

Paper Citation

in Harvard Style

Nyathi T. (2024). Automated Design of a Genetic Algorithm for Image Segmentation Using the Iterated Local Search. In Proceedings of the 16th International Joint Conference on Computational Intelligence - Volume 1: ECTA; ISBN 978-989-758-721-4, SciTePress, pages 189-196. DOI: 10.5220/0012908500003837

in Bibtex Style

@conference{ecta24,
author={Thambo Nyathi},
title={Automated Design of a Genetic Algorithm for Image Segmentation Using the Iterated Local Search},
booktitle={Proceedings of the 16th International Joint Conference on Computational Intelligence - Volume 1: ECTA},
year={2024},
pages={189-196},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012908500003837},
isbn={978-989-758-721-4},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 16th International Joint Conference on Computational Intelligence - Volume 1: ECTA
TI - Automated Design of a Genetic Algorithm for Image Segmentation Using the Iterated Local Search
SN - 978-989-758-721-4
AU - Nyathi T.
PY - 2024
SP - 189
EP - 196
DO - 10.5220/0012908500003837
PB - SciTePress