Face to Face: Anthropometry-Based Interactive Face Shape Modeling Using Model Priors

摘要

This paper presents a new anthropometrics-based methodfor generating realistic, controllable face models. Our methodestablishes an intuitive and efficient interface to facilitateprocedures for interactive 3D face modeling and editing. Ittakes 3D face scans as examples in order to exploit the variationspresented in the real faces of individuals. The systemautomatically learns a model prior from the data-setsof example meshes of facial features using principal componentanalysis (PCA) and uses it to regulate the naturalnessof synthesized faces. For each facial feature, we computea set of anthropometric measurements to parameterize theexample meshes into a measurement space. Using PCA coefficientsas a compact shape representation, we formulatethe face modeling problem in a scattered data interpolationframework which takes the user-specified anthropometricparameters as input. Solving the interpolation problemin a reduced subspace allows us to generate a natural faceshape that satisfies the user-specified constraints. At runtime,the new face shape can be generated at an interactiverate. We demonstrate the utility of our method by presentingseveral applications, including analysis of facial featuresof subjects in different race groups, facial feature transfer,and adapting face models to a particular population group.