Face Recognition Using Eigen-Wavelet-Face Method

Abstract

This work is concerned with investigation for face recognition methods suitable for different environments.  Eigenface method based on Principle Component Analysis (PCA) is modified here by operating on wavelet transformed face image to extract recognition features in a hybrid scheme called Eigen-Wavelet-Face aiming to improve the recognition rate and/or complexity.  Four standard face image databases are used in the work. The databases have different parameters related to size, type, expressions, lighting, orientation, and the number of images per person. The original Eigenface and suggested Discrete Wavelet Transform (DWT) face recognition methods are also used in the work for the sake of comparison. The results showed that the Eigenface method is a time consuming due to its huge computations.  For databases having large number of training images and variations, the proposed hybrid method achieved 100% recognition rate, while for those databases with smaller training sets DWT method obtained the best recognition rate of 95% under favorite condition.

Key words: Face recognition, Eigenface, PCA, DWT, Feature extraction.