Performance Assessment of Principal Component Analysis and Kernel Principal Component Analysis Using TOAM Database

Madandola, Tajudeen Niyi

Department of Computer Sciences, Kwara State College of Education, Oro, Nigeria.

Gbolagade, Kazeem Alagbe *

Department of Computer Science, Kwara State University, Malete, Nigeria.

Yusuf-Asaju Ayisat Wuraola

Department of Computer Science, University of Ilorin, Nigeria.

*Author to whom correspondence should be addressed.


Abstract

Face recognition algorithms can be classified into appearance-based (Linear and Non-Linear Appearance-based) and Model-based Algorithms. Principal Component Analysis (PCA) is an example of Linear Appearance-based which performs a linear dimension reduction while Kernel Principal Component Analysis (KPCA) is an example of non-linear appearance methods. The study focuses on the performance assessment of PCA and KPCA face recognition techniques. The assessment is carried out base on computational time using testing time and recognition accuracy on created database identified as TOAM database. The created database is mainly for this research purpose and it contains 120 face images of 40 persons frontal faces with 3 images of each individual under different lighting, facial expressions, occulations, environment and time. The findings reveal an average testing Time of 1.5475 seconds for PCA and 67.0929 seconds for KPCA indicating a longer Computational time for KPCA than PCA. It also reveals that PCA has 72.5% performance recognition accuracy while KPCA has 80.0% performance recognition accuracy indicating that KPCA outperforms the PCA in terms of recognition accuracy.

Keywords: Kernel principal component analysis, principal component analysis, performance, face recognition, computational time


How to Cite

Niyi, M. T., Alagbe, G. K., & Wuraola, Y.-A. A. (2019). Performance Assessment of Principal Component Analysis and Kernel Principal Component Analysis Using TOAM Database. Asian Journal of Research in Computer Science, 3(2), 1–10. https://doi.org/10.9734/ajrcos/2019/v3i230091

Downloads

Download data is not yet available.