Technology is making possible the creation and handling of multimedia documents which contain video segments. At present there is a lack of tools for automatically indexing and representing these documents. In this paper we restrict our attention to a special class of video documents, those sequences containing images of people in movement. These sequences are analyzed and Spatio-Temporal Indices are extracted and used to describe the body motion. The analysis makes use of a 3-D human model composed of an articulated 10 segment structure. The next step is to recognize the identity of the human in motion. To do this we have developed a face recognition procedure based upon projective invariants. Examples are presented and demonstrate the feasibility of our approach.
Automatic face and gestual recognition for video indexing
IWAFGR 1995, International workshop on Automatic face and gesture recognition, June 1995, Zürich, Switzerland
© INRIA. Personal use of this material is permitted. The definitive version of this paper was published in IWAFGR 1995, International workshop on Automatic face and gesture recognition, June 1995, Zürich, Switzerland and is available at :
PERMALINK : https://www.eurecom.fr/publication/170