Multiple camera based human motion estimation

Akira Utsumi, Hiroki Mori, Jun Ohya, Masahiko Yachida

Research output: Chapter in Book/Report/Conference proceedingConference contribution


We propose a human motion detection method using multipleviewpoint images. We employ a simple elliptic model and a small number of reliable image features detected in multiple-viewpoint images to estimate the pose (position and normal axis) of a human body, where feature extraction is employed based on distance transformation. The COG (center of gravity) position and its distance value are extracted in the process. These features are robust against changes in human shapes caused by hand/leg bending and produce stable pose estimation results. After a pose estimation, a "best view" is selected based on the estimation result and further processing is performed including body-side detection and gesture recognition (in a 2D image of the selected view). This viewpoint selection approach can overcome the problem of self-occlusions. We confirmed the stability of the system through experiments.

Original languageEnglish
Title of host publicationComputer Vision - ACCV 1998 - 3rd Asian Conference on Computer Vision, Proceedings
EditorsRoland Chin, Ting-Chuen Pong
PublisherSpringer Verlag
Number of pages8
ISBN (Print)3540639314, 9783540639312
Publication statusPublished - 1997 Jan 1
Externally publishedYes
Event3rd Asian Conference on Computer Vision, ACCV 1998 - Hong Kong, Hong Kong
Duration: 1998 Jan 81998 Jan 10

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Other3rd Asian Conference on Computer Vision, ACCV 1998
Country/TerritoryHong Kong
CityHong Kong

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)


Dive into the research topics of 'Multiple camera based human motion estimation'. Together they form a unique fingerprint.

Cite this