Exploring the Relationship between Faces and Aging in Image-based Age Transformation

Aging has considerable effects on the appearance of the human face and is difficult to simulate using a universallyapplicable global aging model. In this paper, we focus on the hypothesis that the patterns of age progression (and regression) are related to the face concerned, as the latter implicitly captures the characteristics of gender, ethnic origin, and age group, as well as possibly the lifestyle and health, of the individual. We present a data-driven framework for automatic image-based facial transformation in conjunction with a database of facial images. In addition to a parameterized model for face description, we also build a parameterized model for age-transformation. We introduced and compared several different methods to obtain a person-specific mapping in the model space from an encoded face description to an encoded age-transformation, and we adopted different algorithmic approaches including localization, person-sensitive linear combination, feature-sensitive linear combination, and learning the functional mapping from example transformations using genetic programming. We found that genetic programming is particularly suited for aging simulation where the current understanding is not yet adequate for producing an effective global aging model. In comparison with the existing techniques for age transformation, we have achieved a noticeable improvement in terms of the resemblance between the output images and the actual target images (which are unknown to either the age transformation algorithm or the training process), demonstrating the effectiveness and usability of this new approach.