DocumentCode
602478
Title
3D surfaces representation using geodesic computation and spectral analysis coupled approach: Application to 3D faces
Author
Gadacha, W. ; Ghorbel, Faouzi
Author_Institution
CRISTAL Lab., Nat. Sch. of Comput. Sci. (NSCS), La Manouba, Tunisia
fYear
2013
fDate
20-22 Jan. 2013
Firstpage
1
Lastpage
5
Abstract
In this paper, we intend to introduce a 3D surface representation defined around a given reference point already taken on the surface. Such representation is obtained from a coordinate system which is composed by a set of indexed levels of geodesic curves and radial lines. A sampling criterion through a generalized version of the Shannon theorem allows the determination of the minimum resolution of both curves that describes faithfully the surface. The obtained representation relatively to a unique landmark with the suited resolution is then computed around several reference points to overcome the instability due to landmarks detection errors. Thus, its robustness under small reference points positions distortion is analyzed. Experimentations are performed on the 3D facial database Bosphorus [8]. Results show also the discrimination power of such representation.
Keywords
curve fitting; differential geometry; face recognition; information theory; object detection; solid modelling; spectral analysis; surface structure; 3D faces; 3D facial database Bosphorus; 3D surfaces representation; Shannon theorem; coordinate system; geodesic computation; geodesic curves; indexed levels; landmarks detection errors; radial lines; reference point; sampling criterion; spectral analysis; Databases; Face recognition; Nose; Robustness; Shape; Three-dimensional displays; 3D surface representation; Parametrisation; Shannon theorem; facial surfaces; reference points; robustness;
fLanguage
English
Publisher
ieee
Conference_Titel
Computer Applications Technology (ICCAT), 2013 International Conference on
Conference_Location
Sousse
Print_ISBN
978-1-4673-5284-0
Type
conf
DOI
10.1109/ICCAT.2013.6521957
Filename
6521957
Link To Document