A hybrid method for 3D face localization

Face localization is always a challenging job for a good recognition scheme. If process has to be done for 3D faces, it is quit difficult to localize faces. In this paper we provide a face localization scheme which is based on Average Threshold Value (AVT) and Grayscale Pixel Value Band (GPVB) with the help of Histogram Equalization. We experiment this approach on a synthetic 3D morphable face model of 20 individuals and get 89.51% accuracy among 20 faces with same lighting condition, which is appreciative for further work on same stream.