摘要： 现有虚拟人群运动仿真方法形成的大规模人群动画效率及形成的人群动画效果仿真程度低。针对上述问 题,提出一种将故事板的概念和纹理合成算法相结合的方法,以此形成大规模人群动画。提取视频初始帧中的人群位 置、站位方向、状态和动作信息以及后续帧序列中的人物路径信息,存储于样本故事板。采用基于块的纹理合成方法,利用样本故事板生成大规模故事板,以扩充人群规模。使用大规模故事板中的人群信息数据,结合简单的碰撞检测算法,驱动形成最终的虚拟人群动画。实验结果证明,该方法能够高效地生成效果逼真的大规模虚拟人群动画。

关键词: 基于视频, 故事板, 纹理合成, 数据驱动, 人群仿真, 碰撞检测

Abstract: The existing massive crowd animation simulation methods have the defects of low quality and low efficiency in simulation. In order to solve these two problems,this paper proposes a method of combining storyboard with texture synthesis algorithms which can efficiently generate large scale crowd animation. Crowd’s positions,standing directions, status and information from the first frame including characters ’ path information from the subsequent frames are extracted and stored in the sample storyboard. And combining with block-based texture synthesis method,this sample storyboard can be used to enlarge the size of crowd efficiently. The final large-scale crowd animation can be simulated by using the enlarged storyboard in combination of simple collision detection algorithms. Experimental results prove the high efficiency and quality of this method in generating realistic large-scale virtual crowd animation.

Key words: video-based, storyboard, textural synthesis, data-driven, crowd simulation, collision detection

中图分类号: 

• TP37