Published on December 2021 | Artificial Intelligence

Performance comparison of PCA and ICA algorithms-based face recognition system

Face recognition system has become a dominant research area in biometric studies due to its efficiency and accuracy. This technology has been broadly invested in various security applications for the automatic identification of humans. However, the complexity of human faces representation with the large variation in its characteristics and appearances. This complexity involves adopting powerful algorithms that can effectively learn and overcome such problems with less false results. Many algorithms are proposed for this purpose such as the Principal Component Analysis (PCA) and the Independent Components Analysis (ICA), etc. This work focuses on the implementation of a reliable face recognition system using PCA and ICA as recognition methods and the Euclidean Distance (ED) as a face classifier. A comparison is conducted upon the performances of the PCA and the ICA. These two methods are mainly used in this research for image projection and dimensionality reduction. The classification process is performed by using the distance measure scheme that is adopted by the ED classifier. The comparison is taken for the system robustness evaluation in terms of recognizing a given set of face images.