A novel face recognition approach based on curvelet transform
Jun Liu, Jiyan Huang, Wei Lin, Yineng Zhou, Guocai Mu · 2016
Variant illumination is one of the important factors that affect human face recognition. In the existing study, Retinex algorithm is widely used to mitigate the effect of illumination variations on face recognition, however, the Retinex algorithm is easy to appear the “Halo” phenomenon which leads to the low identification probability. In order to eliminate the “Halo” phenomenon and obtain high recognition rate, we propose a novel face recognition approach, which is based on separate processing on the approximate and detailed components of curvelet transform. Simulations on YaleB database and CMU PIE database show that the proposed method can effectively eliminate the effect of illumination variations and improve the accuracy of face recognition.