3D human motion analysis framework for shape similarity and retrieval
作者：
Highlights：
• A new framework is proposed for 3D human video analysis, based on a new descriptor called Extremal Human Curve.
• Motion analysis is incorporated by employing motion segmentation into clips.
• Clips are represented by trajectories on shape space.
• Clip matching is performed using DTW on the manifold.
• Summarization by clustering is exploited in content-based motion retrieval.
摘要
•A new framework is proposed for 3D human video analysis, based on a new descriptor called Extremal Human Curve.•Motion analysis is incorporated by employing motion segmentation into clips.•Clips are represented by trajectories on shape space.•Clip matching is performed using DTW on the manifold.•Summarization by clustering is exploited in content-based motion retrieval.
论文关键词：Motion analysis,Shape similarity,3D video retrieval,3D human action
论文评审过程：Received 30 March 2013, Revised 3 December 2013, Accepted 30 December 2013, Available online 10 January 2014.
论文官网地址：https://doi.org/10.1016/j.imavis.2013.12.011
