[1]王浩伦,朱业安,徐唯祎,等.步态识别特征的提取和重要性排序[J].中国医学物理学杂志,2019,36(7):811-817.[doi:DOI:10.3969/j.issn.1005-202X.2019.07.013]
WANG Haolun,ZHU Yean,XUWeiyi,et al.Extraction and importance ranking of features for gait recognition[J].Chinese Journal of Medical Physics,2019,36(7):811-817.[doi:DOI:10.3969/j.issn.1005-202X.2019.07.013]
点击复制
步态识别特征的提取和重要性排序()
《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]
- 卷:
-
36卷
- 期数:
-
2019年第7期
- 页码:
-
811-817
- 栏目:
-
医学信号处理与医学仪器
- 出版日期:
-
2019-07-25
文章信息/Info
- Title:
-
Extraction and importance ranking of features for gait recognition
- 文章编号:
-
1005-202X(2019)07-0811-07
- 作者:
-
王浩伦1; 朱业安1; 徐唯祎1; 徐苒1; 黄月姑1; 卢巍2
-
1.华东交通大学交通运输与物流学院,江西南昌330013;2.江西省人民医院康复医学科,江西南昌330006
- Author(s):
-
WANG Haolun1; ZHU Ye’an1; XUWeiyi1; XU Ran1; HUANG Yuegu1; LUWei2
-
1. College of Transportation and Logistics, East China Jiaotong University, Nanchang 330013, China; 2. Department of Rehabilitation
Medicine, Jiangxi Provincial People’s Hospital, Nanchang 330006, China
-
- 关键词:
-
步态障碍; 步态识别; 深度相机; 信息增益; 特征排序
- Keywords:
-
Keywords: gait disorder; gait recognition; depth camera; information gain; feature ranking
- 分类号:
-
R318.01;R742.3
- DOI:
-
DOI:10.3969/j.issn.1005-202X.2019.07.013
- 文献标志码:
-
A
- 摘要:
-
提出一种基于骨骼关键点检测技术的步态识别方法,并讨论常见的步态特征在障碍诊断时的重要性排序,以期为
步态障碍的及时检查和识别提供参考。首先,根据不同的步态模式设立试验组和对照组,受试者按照要求分别完成规定
的范式动作,经由深度相机实时获取受试者骨骼关键点的运动轨迹数据;然后,从获取的数据中提取步态识别特征;最后,
采用以信息增益为依据的模糊二元对比决策方法对步态识别特征的重要度进行排序。通过对每个特征的重要性进行排
序,为步态障碍的诊断和自动识别提供参考。研究结果表明3种步态模式下所有步态识别特征的平均值之间均有显著差
异,特征的优先排序为步频、膝关节最大屈曲角度、足偏角、髋关节最大屈曲角度、步幅、支撑相、步速、髋关节最大伸展角
度、质心左右移动、膝关节最大伸展角度、质心上下移动、步长、步宽。
- Abstract:
-
Abstract: Agait recognition method based on skeleton keypoint detection technology is proposed, and the importance of common
gait features in the diagnosis of gait disorders is discussed, so as to provide a reference for the timely examination and recognition
of gait disorders. Firstly, an experimental group and two control groups are set up according to different gait patterns. All the
subjects complete the prescribed normal motions according to the requirements, and the real-time movement trajectory data of
skeleton keypoints of the subjects are obtained via depth camera. Then, gait recognition features are extracted from the obtained
data. Finally, the importance of gait recognition features is ordered by fuzzy binary comparison decision method based on
information gain, thereby providing a reference for the diagnosis and automatic recognition of gait disorders. Significant
differences are found in all features for gait recognition among 3 different gait patterns. The priority order of gait features is as
follows: stride frequency, maximum flextion angle of knee, toe-out angle, maximum flextion angle of hip, stride, stance phase,
gait speed, maximum extension angle of hip, L/R distance of the center of mass, maximum extenstion angle of knee, U/D distance
of the center of mass, step length and stride width.
备注/Memo
- 备注/Memo:
-
【收稿日期】2019-02-03
【基金项目】国家自然科学基金(61703160);江西省科技厅重点研发
计划(S2019ZPYFB1734)
【作者简介】王浩伦,博士,副教授,研究方向:工业工程、人因工程,Email:
wanghl@ecjtu.edu.cn
【通信作者】卢巍,主任医师,研究方向:偏瘫步态的中西医结合治疗,
E-mail: 13006209911@163.com
更新日期/Last Update:
2019-07-25