多尺度运动片段连接的人体运动动画生成算法  

Human Motion Animation Generation Algorithm Based on Multi-scale Motion Segment

作　　者：蒋巍[1] 卢娜[2] 

机构地区：[1]西安科技大学艺术学院,西安710054 [2]西安航空学院经济管理学院,西安710077

出　　处：《科技通报》2017年第6期170-173,共4页Bulletin of Science and Technology

摘　　要：为了提高人体运动动画生成的有效性,降低动画制作的成本,本文根据聚类以及概率与统计学方法,提出了一种多尺度运动片段连接的人体运动动画生成算法。首先通过建立运动库概率转移和统计模型,挖掘运动数据的时空结构,然后采用多尺度运动数据分析方法实现运动片段连接,能够根据视频内容合成出新的动画数据,从而提高已有运动数据的可重用性,大大降低动画制作成本。实例仿真结果表明,本文提出的方法采用运动数据的无缝连接,使合成的片段在确保平滑过渡的同时仍然保有运动细节,这样动画的效果得到了进一步的保证。In order to improve the effectiveness of human motion animation and reduce the cost of animation production, this paper presents a human motion animation generation algorithm based on multi-scale motion segment according to clustering and probability and statistical methods. First of all,the time and space structure of the motion data is excavated by establishing the probability transfer and statistical model of the motion library. Then, the motion segments are connected by the multi-scale motion data analysis method, and the new animation data can be synthesized according to the video contents, which can improve the reusability of the existing data and greatly reduce the cost of animation.Simulations show that the proposed method is based on the seamless connection of the motion data, so that the synthetic segments can ensure the smoothness of the transition while still keeping the movement details. Thus, the effect of the animation is further guaranteed.

关 键 词：运动片段 多尺度分析 人体动画 动画制作 人体运动库 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]