：虚拟人手操作的逼真性和准确性直接影响到仿真分析结果的可靠性。在虚拟维修作业过程

中虚拟人需要不同的手势对各种工具进行灵活操作，针对该问题提出了从模型中重构手势的思想，

并给出了一种新的关节角提取算法。首先，在人体建模软件中建造需要的目标手势并提取出骨架；

然后用算法对原有手部模型各关节角进行迭代求解，直到该手部模型姿势与目标手势一致，实现对

目标手势的重构。Jack 平台上仿真结果表明，该算法可以快速有效地提取出手部模型的关节-Realistic and accurate virtual hand operation will directly affect simulation validity. For some virtual maintenance

work virtual human might need various hand postures to deftly handle different tools. Aiming at this problem, an approach of

hand posture reconstruction from models was introduced and a new algorithm of extracting hand gesture joint angles was

introduced. Firstly, an expected hand gesture was constructed as the goal gesture in human modeling software and its

skeleton was extracted. Secondly, joint angles of initial hand model were not iteratively adjusted until the adjusted initial

human hand model well matched the goal gesture. Simulation results on Jack platform show this method can effectively

extract hand joint angles for various expected postures.