Age simulation for face recognition

In this paper, an automatic age simulation method used for robust face recognition is proposed. We first use a shape and texture vectors to represent a facial image by projecting it in the eigenspace of shape or texture. Then we use age function combined with aging way classification to estimate age. And we use estimated age, typical vector creating function and the feature vector of the original test image to generate the synthesized feature vectors at target age. At last we reconstruct the shape and texture in eigenspaces and combine them to synthesize facial image at target age. Experiments show that the proposed method can effectively "change" the age of face images, and help to get face recognition result robust to age variation