Illumination Normalization Based on Different Smoothing Filters Quotient Image

The illumination variation problem is still an open question in face recognition in uncontrolled environment. To cope with this problem, many methods are proposed to strengthen illumination compensation and feature enhancement, among which quotient image based methods are reported to be a simple yet practical technique. Recently the SQI, MQI and DMQI are reported to obtain good results in illumination invariant features extracting. However, these techniques can be improved in other ways. In this paper, an effective illumination method is proposed. This method is based on the different smoothing filters and quotient image techniques in analyzing the face illumination. Compared with the traditional approaches: SQI, MQI and DMQI, the experimental results show that our algorithms can significantly improve the performance of face recognition under varying illumination conditions on Yale Face database B.