An Illumination Normalization on Face Images: A Comparative Review

Abstract

Face recognition has attracted significant attention due to its wide range of applications. Recently, researchers have focused on robust face recognition such as face recognition systems invariant to pose, expression and illumination variations. Illumination variation is still a challenging problem in the face recognition research area, especially for appearance-based approaches. This paper will discuss two main categories of the existing illumination methods namely frequency domain approach (DCT & DWT) and filtering approach. This includes the introduction of the representative algorithms, theories, and analyze the advantages and disadvantages of these algorithms correspondingly. The aim of this paper is to explore the technique and limitations of the existing illumination normalization strategies. Based on the reviews that have been conducted, it is found that the filtering technique is more effective when compared to the frequency domain technique. Hence, the implications of the review can help researchers to propose a novel idea in order to solve the existing problems and automatically improve the recognition performance.