Keyframe Extraction Using Linear Rotation Invariant Coordinates

Keyframe extraction is a widely applied remedy for issues faced with 3D motion capture -based computer animation. In this paper, we propose novel keyframe method, where the represented in linear rotation invariant coordinates and dimensions covering 95% of data are automatically selected using principal component analysis. Then, by K-means classification, summarized clustered extracted from each cluster based on cosine similarity. To validate an online user study was conducted. The results show that 45% participants preferred keyframes proposed outperforming alternative 6%.