基于手机放置变化的步态识别

Various placed smartphone based gait recognition

基于手机加速度传感器的步态识别是根据人的生理特点,提取人行走时的加速度步态模式,以区分不同的个体。大多数研究是将加速器固定在同一个位置、同一个方向上,以减少传感器放置变化对识别的影响。文章比较了不同方法,包括统计学和机器学习的方法,用于减少加速器放置变化的影响。而经过滤波、特征提取等处理,使用机器学习的KStar算法分类效果最佳,准确率可达到99.11%,可消减放置变化的影响。

Gait recognition based on accelerometer embedded in smartphone is on the basis of human physiological characteristics,which extracts walking pattern from accelerometer data to distinguish different subjects. In the most published research,in order to reduce influences of changes of sensor placement accelerometers were fixed on the same position and identical orientation. In this paper,we compared different methods including statistics and machine learning to eliminate impacts of changes of accelerometer placement. However,through process of filtering and feature extraction and so on,the KStar algorithm in machine learning classify best and achieved accuracy of 99. 11%,and can eliminate the influences of placement changes.