Facial Feature Points Extraction Technology Research And Its Application In 3d Face Modeling

The need for fast head modeling for realistic facial animation has led researchers to explore 3D model-based algorithms.At present,the head modeling based on the features of human face is more commonly used with three-dimensional modeling of realistic human faces, however,the automatic extraction of feature points still have some inadequacies.In this paper, we develop a practical system for generating a 3D convincing facial model from frontal and profile images taken by an ordinary digital camera.This research is supported by national natural science foundation,the work mainly includes the following aspects:(1) After the study of human cranium and head muscle and according to the standard of MPEG-4,the definition of the facial feature points has been defined to given way to the face contour,facial organ position,shape characterization.(2) By using Selective Masking operator and Fisher's Linear Discriminant to distinguish the regions of facial color and non-facial color.Then fix on facial region,approximate positions of human features were determined by Integral projection and knowledge of facial geometrical structure.At last,the positions of some key feature points were extracted by methods of Hough Transform and Susan operators.(3) With the methods of Susan operators the position of nasal tip has been detected,then with the positions of pupil point and mouth point of the frontal facial,the pupil points and mouth point of profile face were extracted.(4) Utilizes the detected feature points to deform the generic head mesh model such that the deformed model coincides with the detected features.(5) A texture is created by combining the facial textures augmented from the input images,then the synthesized texture mapped onto the deformed generic head model.(6) Based on the theories and researches above,the realistic 3D facial modeling and features extraction system has been implemented by Visual Studio.NET 2003 and OpenGL.