基于运动捕获数据的目标导向运动合成

GOAL-ORIENTED MOTION SYNTHESIS BASED ON MOTION CAPTURED DATA

在计算机动画研究领域,人体运动一直是研究者们感兴趣的内容,逆向运动学是求解人体运动的一项关键技术。用户通过指定相应目标和肢体末端关节,来计算人体全身关节的旋转角,进而得到整体的位姿和运动。传统的数值方法在求解人体运动时没有考虑人体运动的规律,因而得到的结果往往不够真实。提出一种基于真实人体运动数据的实时目标导向运动合成算法,该算法克服了传统数值迭代方法的缺陷,有效改善了单纯使用雅克比矩阵的效果,在动画生成的效率和真实性上都有一定的提高。

In domain of computer animation research, human motion is always the matter attracting the researchers, and the inverse kinematics is a key technique in dealing with the human motion. By assigning the corresponding targets and tile distal end joints of extremity, the users calculate the rotation angles of all joints of human body and then obtain the entire pose Of position, and motion. Traditional numerical methods ignore the law of human motion while dealing with the human motion, therefore the results derived are often less unrealistic. In this paper, we propose a novel real-time goal-oriented motion synthesis algorithm based on actual human motion ＆~ta. This method overcomes the shortcomings of traditional numerical iteration method and effectively improves the results from using the Jacobin matrix only, and has the improvements on efficiency of animation generating and reality to certain point.