Abstract: The reliability of current step counting method generally relies on the body parts where the independent pedometer is equipped, or the positioned location where the mobile phone is set. To solve this problem, this paper proposes a step counting algorithm according to the fixed characteristics of Google glass. The proposed algorithm considers both the acceleration sensor signal and rotation vector sensor signal of Google glass to count steps of pedestrians. In this algorithm, signals are filtered by a fast Fourier transformation filter, and are fitted by a quadratic curve fitting method. Then, the step account is calculated according to the threshold of curve parameter. The experimental results show that the algorithm has a high accuracy of 96%, and can be adapted to some scenarios including ground, slope, and stairs.

Key words: Google glass, pedometer, sensor, curve fitting

摘要： 当前独立的计步器普遍要考虑行人佩戴的部位，如基于手机的计步方法可靠性受限于手机放置位置。根据谷歌眼镜的位置固定特性设计了一种新的计步算法，有效地解决了平地、斜坡、上下楼梯等场景下的计步问题。该算法综合运用谷歌眼镜采集的加速度传感器信号和旋转矢量传感器信号，对这些信号进行快速傅里叶变换滤波处理，用二次曲线拟合信号序列，根据曲线参数阈值进行计步判断。实验结果表明：该算法在平地、斜坡、上下楼梯等场景具有96%以上的准确率。

关键词: 谷歌眼镜, 计步器, 传感器, 曲线拟合