Eigenface vs. Spectroface: A comparison on the face recognition problems

A number of face recognition methods have been proposed. These methods fall into two broad approaches, namely, holistic-based and local-feature-based. Most of holistic-based methods can be classified into two categories, PCA-based and frequency-based categories. This paper introduces a comparison between two holisticbased methods that represent both categories - namely Standard Eigenface method from the PCA-based category and Holistic Fourier Invariant Features (Spectroface) from the frequency-based category. These two methods are tested separately against five main face recognition problems - namely the 3D pose, facial expressions, nonuniform illumination, translation, and scaling - using suitable database(s) for each problem. The results show that the Spectroface method outperforms the Eigenface method in the 3D pose, facial expressions, nonuniform illumination, and translation problems. However, there is no significant difference between both methods in the scaling problem. Finally, in the facial expressions problem, the comparison shows that applying the frequency-based method on the low subband of the wavelet transform is much better than applying the PCA-based method on it.