一种考虑单步统计特征的运动员动作速度识别分析

An analysis of athletes′ movement speed recognition considering the characteristics of single step statistics

为了提高运动员动态下的移动测量能力,设计了一种考虑单步统计特征的运动员动作速度识别方法。先利用脚部传感器采集处于传感器坐标系下的人员运动加速度与角速度,结合加速度信息通过峰值检测方法来完成运动过程的单步划分,得到各单步起始与终止位置,再计算各单步起始位置与终止位置间的角速度与加速度数据便可以得到单步惯性结果。在各分类算法中SVM分类器达到了最高识别准确性,对两种训练集-测试集划分结果进行识别测试都达到了91%以上的准确性。以SVM分类器对单步运动速度进行识别时,可以达到较高的人员行走状态识别准确性。在人员运动速度超过2.5m/s之后,对单步速度产生误判的情况明显多于低速运动状态。

In order to improve the movement measurement ability of athletes under dynamic condition,a method of recognizing athletes′movement speed considering the characteristics of single step statistics is designed.First using the foot sensor acquisition under the sensor coordinate system acceleration and angular velocity,acceleration information through the peak detection method to complete the movement process of single step,and single-step start and end position,then calculate and single-step starting position and end position between the angular velocity and acceleration data can single step inertia results are obtained.Among the classification algorithms in this paper,the SVM classifier has achieved the highest recognition accuracy,and the recognition tests on the division results of the two training set-test sets have achieved the accuracy of more than 91%.When using SVM classifier to identify the single-step motion speed,it can achieve higher accuracy of recognition of walking state.When the movement speed of the personnel exceeds 2.5m/s,the misjudgment of the single-step speed is obviously more than that of the low-speed movement.