提出了一种运动数据重用的方法.采用偏移映射技术构建模型末端的位姿约束轨迹,该轨迹较好地保留了原始运动的特征;利用Unscented卡尔曼滤波方法求解约束,可以实时地生成与原始运动相似的新运动.实验结果表明,该方法可以生成满足不同模型和场景要求的多种运动,扩展了捕获数据的应用范围.
We presented a new motion reuse method. The displacement mapping was used to reconstruct a constrained trajectory for the target character, where the characteristics of the original motion are well preserved. Then based on the constrained trajectory, a real-time unscented Kalman filter solver was applied to generate a new motion similar to the original one. Experimental results show that our method can generate realistic motions suited to different characters and scenes, and consequently extend the reusability of motion-captured data.