摘要
为了有效识别篮球运动员的运动姿态,提升运动员的训练效果,提出了一种基于多特征融合和机器学习的篮球运动姿态识别方法。该方法利用惯性传感器采集运动员的加速度和角速度数据,分别从时域和频域提取多维运动姿态特征,通过特征选择和机器学习实现篮球运动姿态识别。实验结果表明,该方法能够有效识别篮球运动姿态,姿态识别平均准确率达到97%以上。
For effectively recognizing gesture of hoopster and improving training result,a method for recognizing gesture is made out which is based on multi-feature fusion and machine learning.With this method,inertial sensor is used to collect accelerated speed and angular speed data of hoopster.Multi sporting posture feature is got by time domain and frequency domain,and basketball posture is recognized by feature and machine learning.The result shows that basketball posture can be recognized effectively by this method with an accuracy rate of over 97%.
作者
王高宣
Wang Gaoxuan(South China Institute of Software Engineering,Guangzhou University,Guangzhou 510990,China)
出处
《甘肃科学学报》
2019年第3期1-4,11,共5页
Journal of Gansu Sciences
基金
广东高校省级重点平台和重大科研项目(2017GXJK256)
广州大学华软软件学院教育科研项目(ky201730)
关键词
篮球运动
姿态识别
惯性传感器
特征融合
机器学习
Basketball
Posture recognizing
Inertial sensor
Feature fusion
Machine learning
