Face Reconstruction from Skull Shapes and Physical Attributes

摘要: Reconstructing a person's face from its skeletal remains is task that has over many decades fascinated artist and scientist alike. In this paper we treat facial reconstruction as machine learning problem. We use separate statistical shape models to represent the skull morphology. learn relationship between parameters of by fitting them set MR images head using ridge regression on resulting model parameters. Since not uniquely defined shape, allow specify target attributes, such age or weight. Our experiments show results are generally close original face, specifying right attributes perceptual measured difference predicted reduced.