Makeup transfer using multi-example

Abstract

We introduce an individual virtual makeup transfer approach for digital images. Traditional makeup transfer methods focus on the transfer of the whole facial makeup effect, whereas our method can produce high quality individual makeup effect by transferring local makeup effects from multiple examples in different facial component regions. First, we modify the lightness and chroma channels of each facial region of the examples, respectively. Second, we transfer these modified information to corresponding facial region of the target image. Finally, we introduce a hybrid image blending method that allow seamless image composition and combine alpha blending to produce the final result. Experimental results show our method is effective. © 2013 Springer-Verlag.