步态特征提取的K均值聚类自适应判别

Adaptive gait features discrimination based on K means clustering

针对微惯性零速修正算法中步态特征的准确提取,以及步态特征的无规律性成为制约行人导航系统中步态信息提取与辨识的问题,该文提出一种基于K均值聚类自适应的行人步态特征辨别方法。分析行人步态规律并通过设定角速率阈值法对步态特征进行初判后,采用K均值聚类自适应算法设定时间阈值并将误判的步态进行纠正。为验证该算法的普适性,分别针对不同测试个体和同一个体5组不同行走速度条件下的步态特征判别实验,结果表明,本文提出的步态自适应判别方法对不同个体具有良好的适应性;为进一步验证K均值自适应步态判别算法对人员位置解算的准确性,分别开展圆形及400m跑道闭合行走实验,对比不同行走路径对应的位置误差可看出,解算位置误差虽然随行走距离增大而增加,但其相对误差均不超过2%。

According to the accurate extraction of gait features in the micro-inertial zero velocity correction algorithm,and the irregularity of gait features had become a problem that restricted the extraction and identification of gait information in the pedestrian navigation system,apedestrian gait feature discrimination method based on adaptive K-means clustering was proposed in this paper.After the pedestrian gait rule was analyzed and the gait feature was initially judged by setting the angle rate threshold method,the K-means clustering adaptive algorithm was adopted to set the time threshold and the misjudged gait was corrected.In order to verify the universality of this algorithm,the gait feature discrimination experiment was conducted for different test individuals and the same individual with 5 groups of different walking speeds.The results showed that the proposed gait adaptive discrimination method had good adaptability to different individuals.For further verify the accuracy of K-means adaptive gait discrimination algorithm for pedestrian position calculation,the circular and 400-meter runway closed walking experiments were carried out respectively.By comparing the position errors corresponding to different walking paths,it showed that the calculated position error was increased with the walking distance,but the relative error was not more than 2%.