Motion Retargeting for Virtual Human Based on Vector Features of Lower Limbs

Abstract

In view of the contradiction that limited motion data can not meet the requirements of rich and changeable motion, motion reuse technology emerges as the times require, which can drive different human skeleton models with the same motion data to generate corresponding motion. However, when the motion data is used to drive the human models with different heights and bone proportions, motion deformation often occurs. In order to solve this problem, the concept of human lower limb vector is proposed in this paper. By analyzing human motion, it is pointed out that the lower limb vector can maintain the main characteristics of motion, and then a human motion redirection method based on the invariance of lower limb vector feature is proposed. This method can redirect the motion data from the original bone model to the target bone model with different bone length ratio, while maintaining the main characteristics of the original motion. The experimental results show that this method has good motion redirection effect and fast computational efficiency.