计算机与数字工程2025，Vol.53Issue(2)：545-550,6.DOI:10.3969/j.issn.1672-9722.2025.02.043
基于方向特征的DTW动作匹配算法研究
Research on DTW Action Matching Algorithm Based on Direction Feature
葛伟楠 1张明1
作者信息
- 1. 江苏科技大学计算机学院 镇江 212100
- 折叠
摘要
Abstract
Most rehabilitation training systems use depth cameras to obtain 3D coordinates of skeletal points,while 2D images captured by ordinary cameras lack Z-axis coordinates,and the orientation between the human body and the camera has a greater im-pact on the matching of similar actions in 2D images.Aiming at this problem,a new limb angle calculation method with directional features is proposed,which combines pose estimation algorithm to extract skeleton point information,and uses dynamic time warp-ing(DTW)method to calculate the distance between two action sequences.The experimental results show that the matching accura-cy of this method is high,and it can also obtain better recognition results for similar actions in two-dimensional space.关键词
康复训练/方向特征/肢体角度/姿态估计/动态时间规整Key words
rehabilitation training/orientation feature/limb angle/pose estimation/dynamic time warping分类
信息技术与安全科学引用本文复制引用
葛伟楠,张明..基于方向特征的DTW动作匹配算法研究[J].计算机与数字工程,2025,53(2):545-550,6.
