Reconnaissance de gestes en vision par ordinateur

Jérôme Martin 1
1 iMAGIS - Models, Algorithms and Geometry for Computer Generated Image Graphics
GRAVIR - IMAG - Graphisme, Vision et Robotique, Inria Grenoble - Rhône-Alpes
Abstract : This thesis lies in the field of gesture recognition in the context of Human Computer Interaction (HCI). The goal was the design of recognition and understanding systems dedicated to the gestual channel, and their integration into new forms of interaction between users and computerized systems. A review of gestual communication and the domain of HCI allows a reflection of their use for new, natural interaction and the definition of gestures from a designer's point of view. We define a gesture by the concept of its trajectory, which is a parametrized curve over time in characteristics space. Characteristics of a hand are, e.g., its location and its posture. We propose to decompose the recognition problem into three stages: analysis, recognition and interpretation. During the analysis stage we compute the characteristics of a hand for each image of a sequence and create the gesture's trajectory. Spatio-temporal analysis during the recognition step then classifies the gesture between a set of known gestures specific to the application. The interpretation stage finally finds the correspondence between the recognized gesture and the action to be taken by the system. This stage is application dependent. In this thesis we focus on gestures in an intelligent environment, considering gestures for the manipulation of computerized objects and for drawing. We also present an application for human activities recognition based on the movement of individuals in that environment.
Document type :
Theses
Complete list of metadatas

Cited literature [170 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00006749
Contributor : Thèses Imag <>
Submitted on : Tuesday, August 24, 2004 - 5:11:10 PM
Last modification on : Wednesday, April 11, 2018 - 1:54:27 AM
Long-term archiving on : Friday, April 2, 2010 - 8:40:03 PM

Identifiers

  • HAL Id : tel-00006749, version 1

Collections

INRIA | UGA | IMAG

Citation

Jérôme Martin. Reconnaissance de gestes en vision par ordinateur. Interface homme-machine [cs.HC]. Institut National Polytechnique de Grenoble - INPG, 2000. Français. ⟨tel-00006749⟩

Share

Metrics

Record views

1066

Files downloads

5618