A Framework of Local Illumination Normalization for Face Recognition

This paper proposes a framework of illumination normalization algorithm and two new image preprocessing methods according to the framework: piecewise gamma transformation (PGT) and piecewise histogram specification (PHS). They can be used to eliminate the effect of uneven lighting condition effectively and efficiently. The proposed methods have been evaluated based on the Yale face database B and the CMU PIE face database by using PCA and compared with other image preprocessing methods. The experiment results demonstrate that they can significantly improve the recognition performance.