基于加速度传感器的运动步数检测算法研究

为解决人体运动过程中手腕摆动对步数检测干扰的影响，提出了一种区域峰值步数检测算法。该算法根据人体运动时的加速度信号特征，使用高斯滤波去除噪声对加速度信号的影响，基于三轴合加速度数据采用区域峰值检测方法判定有效步伐，同时利用提取的加速度信号幅值的变化量作为特征值，准确实现了稳定状态下的跑步、走路两种动作的自动识别，并根据人体不同的运动状态自动调整采样速率，有效降低了系统功耗。实验结果表明，区域峰值算法相比于动态阀值和单一峰值的算法具有更高的步数检测精确度，可达到95%以上。

Research of Motion Step Detection Algorithm Based on Acceleration Sensor

A new area peak step detection algorithm was presented to decrease the interference of wrist swing to the step dete?ction in the human motion. It was based on the characteristics of the acceleration signal of human motion, using Gaussian filter to cut down the noise effect on acceleration signal, and applying area peak detection to determine the effective step on the basis of the data of triaxial resultant acceleration. Besides, the variation of the amplitude of the acceleration signal was used as the eigenvalue, making realization of the automatic recognition of steady-state walk and run as well as the automatic adjustment of the sampling rate on various motional state, effectively reducing power dissipation. The experimental results show that, compared with dynamic threshold algorithm and single peak algorithm, the area peak step detection algorithm has highly precision in step detection, up to more than 95%.