基于手机加速度传感器的高精低采样计步算法设计  

A high-accuracy and low-sampling-rate step counting algorithm design based on smartphone's accelerometer

作　　者：梁久祯[1] 朱向军[1] 陈璟[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《西北大学学报（自然科学版）》2015年第5期738-744,共7页Journal of Northwest University（Natural Science Edition）

基　　金：国家自然科学基金资助项目(61170121)

摘　　要：针对传统的计步算法在计步准确度和算法复杂度上难以达到良好平衡的问题,提出了一种双算法计步模型:辅助计步算法采用信号幅值分析的方法,避免了传统的幅值检测或过零检测过程带来的误差问题;主体计步算法借助于阈值校正因子,在保证较高计步精度的同时,降低了算法的复杂度。计步模型的运算过程与手机摆放位置无关,并且能够很好地适应不同用户的步态习惯。实验结果表明,即使在低频率(10Hz)采样时,其平均计步准确度也可以达到90%以上。For the tradeoff between the algorithm＇ s accuracy and complexity existing in step detection techniques,a dual-algorithm based on step-counting model is proposed. An analysis of signal amplitude is adopted in the auxiliary part,which avoids the error of amplitude comparison or zero crossing. By using the threshold calibration factor provided by the auxiliary part,the primary part reduces the complexity of the algorithm,while ensuring the high accuracy of it. The operation process of the model doesn＇t depend on the placement of the smartphone,and could adapt to different users＇ walking pattern. The experiment demonstrates that the algorithm can reach an average accuracy above 90% even the sampling frequency is 10 Hz.

关 键 词：计步算法 加速度传感器 信号幅值分析 低采样 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]