Title :
Modelling faces dynamically across views and over time
Author :
Li, Yongmin ; Gong, Shaogang ; Liddell, Heather
Author_Institution :
Dept. of Comput. Sci., London Univ., UK
Abstract :
A comprehensive novel multi-view dynamic face model is presented in this paper to address two challenging problems in face recognition and facial analysis: modelling faces with large pose variation and modelling faces dynamically in video sequences. The model consists of a sparse 3D shape model learnt from 2D images, a shape-and-pose-free texture model, and an affine geometrical model. Model fitting is performed by optimising (1) a global fitting criterion on the overall face appearance while it changes across views and over time, (2) a local fitting criterion on a set of landmarks, and (3) a temporal fitting criterion between successive frames in a video sequence. By temporally estimating the model parameters over a sequence input, the identity and geometrical information of a face is extracted separately. The former is crucial to face recognition and facial analysis. The latter is used to aid tracking and aligning faces. We demonstrate the results of successfully applying this model on faces with large variation of pose and expression over time
Keywords :
computational geometry; face recognition; image sequences; parameter estimation; affine geometrical model; face appearance; face recognition; faces modelling; facial analysis; global fitting criterion; large pose variation; model parameters estimation; multi-view dynamic face model; shape-and-pose-free texture model; sparse 3D shape model; video sequence; video sequences; Active appearance model; Active shape model; Aging; Computer science; Data mining; Face recognition; Image sequence analysis; Solid modeling; Support vector machines; Video sequences;
Conference_Titel :
Computer Vision, 2001. ICCV 2001. Proceedings. Eighth IEEE International Conference on
Conference_Location :
Vancouver, BC
Print_ISBN :
0-7695-1143-0
DOI :
10.1109/ICCV.2001.937565
