提出了一种新的重新定向手势运动数据的方法.它把通过运动捕获设备捕获的手势数据重新定向到具有不同大小和比例的虚拟人模型上,从而生成与原运动含义相同且真实自然的动画.首先通过手势特征分析在人体上选择关键敏感点和次关键敏感点,然后利用提出的基于空间相对位置的方法确定重定向后敏感点的位置,最后采用改进的逆运动学方法实现了手势运动的重定向.实验表明,提出的方法使得中国手语合成系统的可懂度提高30%以上.
Existed motion retargeting algorithms mostly focused on the motion between the character and the environment and seldom considered the interaction among the segments of a character, especially for the hand gesture. In this paper a new method for motion retargeting for the hand gesture is proposed. Realistic and natural animations with similar meanings to the original can be produced through the way. First many sensitive points are defined on the human body, and key sensitive points and secondary sensitive points are selected through analyzing the importance of the sensitive points. Then a novel method based on the relative position of the original sensitive points is proposed to compute the target sensitive point position, Finally an adapted IK solver is utilized to realize the retargeting problem. Experimental results show that the proposed method dramatically improves the recognition rate of the Chinese sign language synthesis system about 30%.