Margin Refine of Candide-3 Model for Facial Deformation

Abstract:

Because of the structure of the CANDIDE model is simple and the quantity of its marginal gridlines is small, an obvious marginal trace would appear while the face changed. In view of this, this paper proposed a method which refined the facial outline gridlines to improve the effect of facial transformation. Firstly, the feature point of the pupil is set to enhance the location veracity of the model, then 44 key feature points are chosen to represent the information of the basic facial features. The face matching is realized by ASM algorithm, then fit and refine the extracted contour line personalized to enhance the expressive force of the model of CANDIDE-3. Experimental results show that the result of facial transformation has been optimized. The effect of the transformation is optimized on the premise of the computation speed is guaranteed, and the result looks more naturally.

• Periodical: Advanced Materials Research (Volumes 926-930)
• Pages: 3073-3078
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.926-930.3073
• Online since: May 2014
• Authors: Xu Xu Chen, Hui Jian Han*
• Keywords: CANDIDE, Feature Points, Outline Gridlines Fit, Refine

References

[1] J.L. Chang, A study on the expression synthesis for 3D face, University of Science and Technology of China, Anhui, China, (2009).

[2] Y.J. Wang, Face recognition method based on local matching. Nanchang University, Nanchang, China, (2012).

[3] H. Duan, Y.X. Wang, Y.M. Cheng, S.S. Cai, Drive muscle face model using single video, Journal Of System Simulation, 2002(9), 1268-262.

[4] Y.P. Jiang, X.J. Wang, Y.J. Liu, X.F. Zhang, W.P. Tai, Research of face depth generation information based on candied model, Computer Technology And Development, 2012(4), 93-100.

[5] A hlberg J. CANDIDE-3—an updated parameterized Face. Report No. LiTH-ISY-R-2326, Sweden: Dept of Electrical Engineering, Linkoping University, (2001).

[6] F.S. Hu, Y.P. Lin, B.J. Zou et al, 3D Face Reconstruction Based on Candide-3 for Face Recognition. China The 2008 Inter-national Congress on Image and Signal Processing CISP2008(4): 158.

DOI: 10.1109/cisp.2008.345

[7] X.J. Du, Y.Z. Bai, Y.Q. He, Study of pose and expression variant face recognition based on candied-3 modle. Computer Engineering And Design, 2012(3), 1017-1021.

[8] Y.Q. Xiao, J. Hou, X. Wang, Self-adaption 3D head gridding model study based on frontal facial photo, Computer Simulation, 2013(7), 412-426.

[9] C.Z. Sun, Research and application of key feature point locating face, University of Electronic Science and Technology of China, Chengdu, China, (2009).

[10] C.H. Du, J. Yang, Q. Wu, T.H. Zhang, H. H Wang, L. Chen, Z. Wu, Extended fitting methods of active shape model for the location of facial feature points, Lecture Notes in Computer Science, Berlin Heidelberg: Springer-Verlag, 2006: 610-618.

DOI: 10.1007/11949619_54

[11] Y. Tian, 3D Face reconstruction based on a single frontal face image, Chongqing University, Chongqing, China, April, (2011).