HCI 2003, 10th International Conference on Human Computer Interaction, June 21st-27th, 2003, Crete, Grece
Complete interaction amongst humans and machines unavoidably needs computers to understand human emotions. Most emotive information comes from facial motion and expression. This article presents the design of a new procedure for image analysis that is able to understand facial actions on monocular video sequences, without imposing restrictions on the speaker or its environment. The exposed technique follows a global-to-specific analysis approach that tries to imitate the way people analyze face motion: by dividing this analysis in processes of different level of detail.
Type:
Conference
Date:
2003-06-21
Department:
Digital Security
Eurecom Ref:
1208
Copyright:
© LEA. Personal use of this material is permitted. The definitive version of this paper was published in HCI 2003, 10th International Conference on Human Computer Interaction, June 21st-27th, 2003, Crete, Grece and is available at :
