Automatic facial image standardization based on active appearance model

Image standardization can benefit the facial image in many applications, such as face recognition, virtual reality, pose rectification and so on. However, as far as we know, there is no automatic facial image standardization method proposed until now. In this paper, we have developed an automatic facial image standardization algorithm based on Active Appearance Model (AAM). The right and left eyes' coordinates are accurately extracted by AAM, and used to compute the displacement, by which an affine transformation model is estimated to translate the original image into a standard image. The face in the standard image is located at a typical coordinate with standard size. Experimental results demonstrate that the proposed method could provide an accurate and reliable way for facial image standardization.