Support vector machines for face recognition with two-layer generated virtual data

This paper presents support vector machines (SVM) for few samples-based face recognition with two-layer artificially generated virtual training data. The few samples cannot express all the conditions of the test data. Thus, we generalize the samples and the feature data to other conditions according to the distribution. First, correspond to the original face images, by locating the eyes center on the face images and facemask template; second is to the feature vectors, we get the feature data by principal component analysis to the face images, then use linear interpolate and extrapolate methods to generate new data. After all the data drawn, SVM is used to train and test. In the ICT-YCNC face database, the proposed system obtains competitive results, and shows the methods are available.