A Study on Generating and Matching Facial Control Point Using Radial Basis Function

Abstract

3D face animation has been applied widely through virtual reality, MPEG-4 image compression, teleconference, advertisement, and game. Especially, it has developed to a point where it can express 3D face in real time thanks to a development of hardware and software. General 3D face modeling generates 3D face model and matches face model by using vertex points on 3D mesh as control points or generating random control points to match the image corresponding to each individual's face form. However, since this type of work uses control point coordinates on a whole mesh, it takes a lot of time and effort to distort mesh form or to match the image corresponding to user's face form. Therefore, this study developed a method that generates 3D individual face model automatically using 2D full face image. In addition, this study provided a face control point generating and matching method using FFPs(Facial Feature Points) information extracting method using RBF from 3D face mesh.