期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：1
页码：327-336
出版社：Journal of Theoretical and Applied
摘要：One of the main complications in face recognition applications, it is non-linearity. Support vector machine is one of the most significant classification techniques in last a few years which can determine the global finest solutions in many complicated problems with minor number of training samples. However, selecting the ideal parameters for SVM is a major challenge especially when SVM used in face recognition applications. Numerous methodologies are utilized to manage this issue, for example, PSO, OPSO, AAPSO and AOPSO. Nevertheless, there is a room of upgrades still exists respects this sort of enhancement process. Recently, an enhanced version of PSO has been introduced called Median-oriented PSO (MPSO) with a few favorable benefits: simple to execute, insensitive to variable dimension, and no requirement for any calculation particular parameters. In this study, a new face recognition technique based on a combination of Median-oriented particle swarm optimization and support vector machine is proposed. The proposed scheme is called (MPSO-SVM) and we introduced it as a face recognition technique. In MPSO-SVM, MPSO is utilized to discover the optimal parameters of SVM. Two human face datasets: SCface dataset and CASIAV5 face dataset are used as a part of the experimentation to assess the proposed MPSO-SVM in recognizing the human faces. The proposed technique is compared with PSO-SVM, OPSO-SVM and AAPSO-SVM and the results showed that the proposed MPSO-SVM has higher face recognition accuracy than the other approaches.
