Abstract

In this paper, a novel method of human motion new style generated by non-linear subspace is proposed. Due to high dimensionality of Motion's features, the Isomap nonlinear dimension reduction analysis is used. An algorithmic framework based on parameters decomposing and extraction is used to approximate the optimal mapping function for handling new data. Then Synthesis is applied in such new low-dimensional subspaces and new motion styles can be control and reconstructed. Experimental results show that our methods are effective.

Paper Details