基于余弦距离的人体运动数据行为分割算法

来源期刊:中南大学学报(自然科学版)2014年第4期

论文作者:邢薇薇 仝磊鸣 张毅 任程

文章页码:1128 - 1137

关键词:人体运动分割;运动捕捉数据;骨骼夹角;余弦距离;曲线简化

Key words:human motion segmentation; motion capture data; joint angle; cosine distance; curve simplification

摘    要:对人体运动捕捉数据进行行为分割是人体运动数据分析与合成中的关键处理步骤,为此,提出一种新的人体运动数据行为分割算法。采用骨骼夹角直方图刻画人体运动统计特征,使用余弦相似度作为人体运动数据骨骼夹角直方图特征的相似性度量,实现对运动行为的自动分割。对于给定的人体运动捕捉序列,首先定义滑动比较窗口,计算当前窗口范围内运动序列前、后2部分骨骼夹角直方图统计特征的余弦相似度,然后通过在运动序列上滑动该窗口,获得运动序列的余弦相似度曲线,曲线最小值位置即为不同类型行为的分割点。在CMU人体运动捕捉数据库上进行数值实验。研究结果表明:本文算法能够实现对人体运动捕捉数据的自动行为分割;与广泛采用的基于PPCA的行为分割方法相比,本文算法具有良好的性能。

Abstract: Considering that the human motion segmentation is one important process for human motion data analysis and synthesis, a novel segmentation algorithm for motion capture data was proposed to segment motions into distinct behaviors. The method was based on the assumption that motions with the same type had similar histogram of angle between bones. Angle histogram was employed to represent the human motion, and cosine distance was utilized to measure the similarity of motions represented by angle histograms. A motion sequence was given, and a sliding window which moved from the starting frame to the end was firstly defined. In each window, the cosine similarity between the first half window and the second half was computed, and so a cosine distance curve with the motion was obtained. The minimums of the curve identified the cut between different types of motion behaviors. The present method was tested on the Carnegie Mellon Motion Capture database. The results show that the method can achieve the automatic segmentation for human motion capture data, and has good motion behavior segmentation performance compared with the classical PPCA based segmentation algorithm.

相关论文

  • 暂无!

相关知识点

  • 暂无!

有色金属在线官网  |   会议  |   在线投稿  |   购买纸书  |   科技图书馆

中南大学出版社 技术支持 版权声明   电话:0731-88830515 88830516   传真:0731-88710482   Email:administrator@cnnmol.com

互联网出版许可证:(署)网出证(京)字第342号   京ICP备17050991号-6      京公网安备11010802042557号