Base mesh extraction for different 3D faces based on ellipse fitting

Extracting base mesh for different 3D faces is a crucial and fundamental problem in 3D facial correspondence research. Establishing 3D facial correspondence can contribute to many computer graphics tasks and applications such as statistical 3D face model building, facial animation, and facial texture transferring, etc, but it suffers from the fact that different 3D faces usually have inconsistent boundaries and irregular representation. In this paper, we propose a novel method to extract 3D face base mesh for different 3D faces. The proposed method is fully automatic, and its main observation is that the ellipsoidal area around the five sense organs of 3D faces is the consistently matched area between different 3D faces. To extract the ellipsoidal area, we present two main approaches: (1) locating elliptical boundary points, (2) ellipse fitting and base mesh extraction. In the first approach, we present ways to determine the facial midline and the iso-curves which are parallel and perpendicular to the facial midline. During analyzing the facial midline and the iso-curves, we define a modified resultant force and its signed resultant force strength, then use them to locate the positions of a series of elliptical boundary points. In the second approach, we use these elliptical boundary points to fit an ellipse in the 2D representation of the 3D face scan by solving an over determined system of equations in the least square sense. Finally we use the fitted ellipse to extract the 3D face base mesh. We evaluate our method and compare it with other methods which are mostly related to ours in a large scale publicly available 3D face data set, BJUT-3D face database. The experimental results show that our method is robust and effective, and achieves better performance than existing methods.