A STUDY OF SYNTHESIZING NEW HUMAN MOTIONS FROM SAMPLED MOTIONS USING TENSOR DECOMPOSITION (WedAmOR1)

Author(s) : Rovshan Kalanov (Waseda University, Japan) Jieun Cho (Waseda University, Japan) Jun Ohya (Waseda University, Japan)

Abstract : This paper applies the algorithm to a new synthesis appli-cation: using sampled human motions of some ages and emotions, new human motions of other age and emotions are synthesized. Human motion is the composite conse-quence of multiple elements, including the action per-formed and a motion signature that captures the distinc-tive pattern of movement of a particular individual. We applied an algorithm, based on Tensor Decomposition, which is capable of extracting these motion elements and recombining them in novel ways. The algorithm analyzes motion data yields a generative motion model that can synthesize new motions in the distinctive styles of these individuals.

Menu