基于中心距离特征的人体运动序列关键帧提取

Key Frame Extraction Using Central Distance Feature for Human Motion Data

运动捕获数据的关键帧是原始运动序列的简洁表示,对于运动压缩、运动检索和运动分割起着重要的作用。提出了一种基于中心距离特征的人体运动捕获数据关键帧提取方法,通过提取四肢到中心点ROOT的距离,得到一组中心距离特征,将特征分为上肢和下肢来分别表示,并提取上下肢的距离模,得到二维的特征向量模;然后采用主成分分析得到一维特征,并提取其局部极值点作为初始关键帧;最后通过对初始关键帧的重新筛选与插入得到最终关键帧序列。实验结果表明,该方法提取的关键帧序列在视觉上能够很好的概括原始运动序列的内容,且具有高压缩率。

The key frames of the motion capture data are the concise description for the original motion sequence,which play an important role for the motion compression,motion retrieval and motion segmentation.A key frame extraction method based on the central distance features for the human motion capture data was proposed.The approach was divided into three main stages.In the first stage,a set of central distance features from the center joint ROOT to limbs was selected,and those features were divided into the upper and lower limbs norms.In the second stage,PCA method was used to get the one dimension principal component,which was used to extract the local optimum as the initial key frames.In the last stage,the initial key frames were filtered and inserted to get the final key frames.Experimental results show that the proposed method can get the needed key frames,which can have good visual generalization of the original motion sequence,and also be of high compression ratio as well.