混合运动模式下的双重阈值零速区间检测算法

Double-threshold zero-velocity interval detection algorithm for multi-movement patterns

针对固定阈值零速检测算法在混合运动模式下容易解算出错等问题,在分析了正常行走、上楼和下楼等三种室内常见运动模式的基础上,提出了一种混合运动模式下的双重阈值零速检测算法。该算法首先采用角速度变化特征点识别出正常行走、上楼和下楼等三种运动模式,然后根据不同的运动模式匹配相应的加速度和角速度阈值以及时间窗口的大小,利用双重阈值配合时间窗口的方法检测得到零速区间。最后,采用该算法对不同的行人的步态在混合运动模式下进行了检测,结果表明该方法可有效识别每步的运动模式,并且能够完全检测出每步的零速区间。在定位结果上相比于固定阈值零速检测法,该算法的定位精度平均提升了70%以上。

To deal with the low accuracy of zero velocity detection in multi movement patterns based on fixed threshold,a zero-velocity interval detection algorithm with double thresholds is proposed,which is suitable for multi movement patterns based on the analysis of pedestrian's three indoor movement patterns,i.e. walking,upstairs and downstairs.First,this algorithm identifies the movement pattern (walking,upstairs,or downstairs)using feature points of angular velocity.Then,according to different motion patterns,the corresponding acceleration and angular velocity thresholds as well as the size of time window are matched, and the zero velocity interval is detected by means of double thresholds and time window.The detection tests of different pedestrians'multi movement patterns by the proposed algorithm show that the movement pattern of each step can be identified,and the zero velocity intervals can be detected.Compared with fixed threshold zero-velocity detection,the position accuracy of the proposed algorithm can be improved by over 70%.