Animation synthesis by observation and learning

Kohta Aoki, Ken'Ichi Morooka, Osamu Hasegawa, Hiroshi Nagahashi

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

Abstract

This paper introduces a method for the animation of things based on the observation of natural phenomena and on the synthesis of their behavioral patterns using machine learning methods. The natural phenomena to be animated is recorded using a video camera, and its characteristics behavior is captured. A data sequence representing the subject behavior is obtained from the captured video. By learning the inherent structure in the feature space of some sample data, the learned model can synthesize a novel data sequence from the existing sequences. The generated sequences of behavioral patterns could differ from every original data sequence but preserve characteristics of the subject behavior. We demonstrate the natural animation synthesis through such behavioral pattern sequences, and produce some realistic animation which depict the subject.

Original languageEnglish
Title of host publicationProceedings - 2003 IEEE International Symposium on Computational Intelligence in Robotics and Automation
Subtitle of host publicationComputational Intelligence in Robotics and Automation for the New Millennium
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1258-1263
Number of pages6
ISBN (Electronic)0780378660
DOIs
Publication statusPublished - 2003
Event2003 IEEE International Symposium on Computational Intelligence in Robotics and Automation, CIRA 2003 - Kobe, Japan
Duration: Jul 16 2003Jul 20 2003

Publication series

NameProceedings of IEEE International Symposium on Computational Intelligence in Robotics and Automation, CIRA
Volume3

Other

Other2003 IEEE International Symposium on Computational Intelligence in Robotics and Automation, CIRA 2003
CountryJapan
CityKobe
Period7/16/037/20/03

ASJC Scopus subject areas

  • Computational Mathematics

Fingerprint Dive into the research topics of 'Animation synthesis by observation and learning'. Together they form a unique fingerprint.

Cite this