Abstract: MEMS inertial sensors applied to human motion capture, but because of the MEMS sensor system errors, the virtual people cannot operate precisely and stably. In order to achieve tracking the movement of an arm accurately, by analyzing the movement of a virtual arm, measuring the distance between the finger of the arm and the center of the object in virtual scene, combined with the virtual arm movement characteristics, a composite Kalman motion estimation method was designed. The model divided arm movement into a uniform model, a uniformly accelerated model and several typical phase models. Depending on the different distance measure, used the different models to make an accurate estimate of the movement of a virtual arm. Through experiments, the composite Kalman motion estimation method could estimate virtual arms in a virtual scene accurately.

Key words: composite model, kalman motion estimation, virtual scene, distance measure