Study Of 3D Facial Profile Extraction

With the development of the society, requests for fast and effective auto status identification are increasingly urgent. Among them, using face features is both the most natural and the most direct means. Auto facial metrology has important applied value in face detection and recognition, realistic face modeling and animation, facial estimation of gesture and illumination, etc. As a result, auto facial metrology is the research focus in the world.This thesis gives the future application of auto facial metrology and the present research, a 3D Chinese color face database is constructed by Beijing university of Technology. Utilizing 3D data in the database, a method of reverting 3D face data is presented. And an algorithm of facial profile extraction and analysis is put forward.1. The reverting processing of 3D face image data. Based on understanding the data structure fully in the 3D database the topic brings forward a new algorithm which is to revert face data to the state which 3D faces should not be standardized, the bidirectional mapping between the shape of 3D face and the texture of 2D face is constructed. And 3D face data are regulated for coordinates in an appropriate range.2. Facial profile extraction and analysis. There is a new technique presented for detecting nose ridge points to locate the mid-line of the face. The result is an aligned 3D head with the line of best fit for the nose ridge to produce the facial profile. Utilizing tangency-based techniques where lines tangent to the profile are used to locate fiducials relative to other fiducials. There are eleven facial profile fiducials that are most often used in facial profile analysis. The experiment results of facial profile extraction and analysis are presented and analyzed.