Image Segmentation as a Multiobjective Optimization Problem

Diego Oliva, Mohamed Abd Elaziz, Salvador Hinojosa

Research output: Chapter in Book/Report/Conference proceedingChapter

2 Citations (Scopus)

Abstract

This chapter presents an alternative multiobjective image segmentation algorithm which aims to find the suitable solutions that balance between two different objectives. The proposed method depends on the improvement of the ability of the multi-verse optimization algorithm using the opposite based learning method. The proposed approach, called MOGWO, uses the Otsu and Kapur function to determine the approximate Pareto-optimal set of solutions. There are a set of experiments are performed using seven images to evaluate the performance of the proposed method as a multiobjective segmentation method. In addition, it is compared with other multiobjective meta-heuristics such as NSGA-II, MOPSO, and MOEAD. The comparison used the PSNR and SSIM to evaluate the segmented images and Hypervolume to assess the solutions. The experimental results show that the proposed method outperforms the other multiobjective algorithms based on the performance measures.

Original languageEnglish
Title of host publicationStudies in Computational Intelligence
PublisherSpringer Verlag
Pages157-179
Number of pages23
DOIs
Publication statusPublished - 2019
Externally publishedYes

Publication series

NameStudies in Computational Intelligence
Volume825
ISSN (Print)1860-949X

ASJC Scopus subject areas

  • Artificial Intelligence

Fingerprint Dive into the research topics of 'Image Segmentation as a Multiobjective Optimization Problem'. Together they form a unique fingerprint.

Cite this