Manuscript received March 14, 2025; revised April 17, 2025; accepted June 16, 2025; published October 17, 2025.

Abstract—Face recognition plays a crucial role in our daily lives by identifying and authenticating individuals. One of the most widely used methods in this domain is the Support Vector Machine (SVM), a supervised machine learning classifier. However, optimizing SVM parameters is a key challenge. This study proposes a comparative evaluation of two optimization algorithms—Particle Swarm Optimization (PSO) and Adaptive-Opposition PSO (AOPSO)—for enhancing SVM performance in face recognition tasks. The proposed methods, PSO-SVM and AOPSO-SVM, were implemented and tested on two benchmark face datasets: CASIA V5 and FEI. Using 10-fold cross-validation, the models were evaluated based on classification accuracy, computational time, and optimization performance. The experimental results show that AOPSO-SVM consistently outperforms the standard PSO-SVM model. Specifically, AOPSO-SVM achieved an accuracy of up to 91.4% on the CASIA V5 dataset and 80.1% on the FEI dataset, while also reducing computational time and improving convergence behavior. These results demonstrate the effectiveness of AOPSO in optimizing SVM parameters for robust and efficient face recognition.

Keywords—optimization, support vector machine, particle swarm optimization, face recognition

Cite: Asaad Noori Hashim, Safaa Jasim Mosa, Abbas F. H. Alharan, and Ahmed J. Obaid, "Evaluation Study for the Performance of PSO and AOPSO Algorithms in Optimizing the SVM Classifier," Journal of Image and Graphics, Vol. 13, No. 5, pp. 521-527, 2025.

Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC-BY-4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.