Human Motion Synthesis based on Iterated Function Systems

Stephen Wang-Cheung Lam and Powis Lai-Yin Leung

Department of Computing, The Hong Kong Polytechnic University, Hung Hom, Kowloon.

Email: cswclam@comp.polyu.edu.hk Fax: 2774 0842 Tel: 2766 7301

 

Abstract

This paper describes a novel approach to automatic motion sequence generation. The basic algorithm underlying our approach is the iterated function systems (IFS) which have already found many applications in image compression and pattern generation. In our system, a set of key-frames are first specified by the user. Subsequently, IFS based interpolation scheme is employed to generate a sequence of motions having self-similar characteristics. Our ultimate purpose is to produce realistic repetitive motion, which possesses a main theme but the details varies over time. Our algorithm can find applications in generating human motion like dancing or walking which involves long sequence of repetitive movements of a puppets limbs and body.

Keywords: Iterated function systems, puppet, motion.