Skip to Main content Skip to Navigation
Theses

Contribution à l'analyse et à l'interprétation du mouvement humain: application à la reconnaissance de postures

Abstract : This Ph.D. thesis research work is dedicated to the analysis and the interpretation of human motion with an application to posture recognition. Human motion analysis and interpretation in computer vision have numerous applications domains such as videosurveillance, mixed-reality applications and advanced man-machine interfaces. We propose here a real-time system that allows human motion analysis and interpretation.

Human motion analysis includes several processing steps of image processing such as segmentation of moving objects, temporal tracking, skin detection, human body models, and actions or pose recognition. We propose a temporal tracking method in two stages that allows to track one or several persons even if they occlude each other. This method is based on the computation of bounding boxes overlap and a partial Kalman filtering. Then we explicit a skin detection method by a color approach in order to localize their faces and hands. All these preliminary steps give access to a lot of low-level data. In a last part, we use some of these data to perform static human body posture recognition of people among the four following postures: standing, sitting, squatting and lying. Several results illustrate the advantages and limitations of the proposed methods, as their efficiency and robustness.
Complete list of metadatas

Cited literature [238 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00156572
Contributor : Vincent Girondel <>
Submitted on : Thursday, June 21, 2007 - 3:38:07 PM
Last modification on : Friday, November 6, 2020 - 4:10:28 AM
Long-term archiving on: : Monday, September 24, 2012 - 10:15:26 AM

Identifiers

  • HAL Id : tel-00156572, version 1

Collections

Citation

Vincent Girondel. Contribution à l'analyse et à l'interprétation du mouvement humain: application à la reconnaissance de postures. Interface homme-machine [cs.HC]. Institut National Polytechnique de Grenoble - INPG, 2006. Français. ⟨tel-00156572⟩

Share

Metrics

Record views

590

Files downloads

613