Statistical Evaluation of Enhanced Face Recognition Techniques under Variable Constraints
Abstract
Face recognition is a dedicated process in the brain which starts from 24 hours of birth. The idea of mimicking this essential skill in human beings by machines can be worthwhile though developing an intelligent and self-learning system may require adequate supply of information to the machine. This study modified Principal Component Analysis and Singular Value Decomposition (PCA/SVD) face recognition algorithm by incorporating Fast Fourier Transform (FFT) and Whitening at preprocessing stages. This study proposed multivariate statistical evaluation of the recognition performance of these face recognition algorithms under varying constraints. In particular, the Repeated Measures Design, Paired Comparison test and Profile Analysis were used for performance evaluation of the algorithms on the merit of effciency and consistency in recognizing face images with variable facial expressions. The results indicated that, FFT-PCA/SVD is more consistent and computationally efficient when compared to PCA/SVD and Whitened PCA/SVD. Fast Fourier Transform is recommended as a viable noise removal mechanism, which should be adopted during preprocessing stages in image or pattern recognition.
Description
A thesis submitted to the Department of Mathematics,
Kwame Nkrumah University of Science and Technology In
partial fulfillment of the requirement for the Degree of Doctor of Philosophy,