Tracking hybrid 2D-3D human models from multiple views

Author

Ong, Eng-Jon ; Gong, Shaogang

Author_Institution

Dept. of Comput. Sci., Queen Mary & Westfield Coll., London, UK

fYear

1999

fDate

1999

Firstpage

11

Lastpage

18

Abstract

A novel framework is proposed under which robust matching and tracking of a 3D skeleton model of a human body from multiple views can be performed We propose a method for measuring the ambiguity of 2D measurements provided by each view. The ambiguity measurement is then used for selecting the best view for the most accurate match and tracking. A hybrid 2D-3D representation is chosen for modelling human body poses. The hybrid model is learnt using hierarchical principal component analysis. The CONDENSATION algorithm is used to robustly track and match 3D skeleton models in individual views

Keywords

gesture recognition; image representation; principal component analysis; tracking; 3D skeleton model; CONDENSATION algorithm; ambiguity measurement; human body; hybrid 2D-3D representation; multiple views; principal component analysis; robust matching; tracking; Biological system modeling; Bones; Computer science; Educational institutions; Humans; Kinematics; Parameter estimation; Principal component analysis; Skeleton; Solids;

fLanguage

English

Publisher

ieee

Conference_Titel

Modelling People, 1999. Proceedings. IEEE International Workshop on

Conference_Location

Kerkyra

Print_ISBN

0-7695-0362-4

Type

conf

DOI

10.1109/PEOPLE.1999.798341

Filename

798341