Algorithm of Preparation of Training Sample Using 3d-Modelling of Faces

Abstract

The algorithm to preprocess the training sample of multi-class classifier support vector machine (SVM) is described in the paper. This provided approach is based on the modeling of possible changes of the face features of human recognition, which influence on training classifier and respectively on further results of recognition. Age, emotional face expressions, head rotation, different lighting conditions, noise, and some combinations of mentioned parameters were selected as the major parameters for modeling.