基于三轴加速度传感器的人体日常体力活动识别系统设计  

Recognition System of Human Daily Physical Activity Based on a 3D Acceleration Sensor

作　　者：李丹[1,2] 陈焱焱[2] 姚志明[2] 杨慧亮[1,2] 

机构地区：[1]中国科学技术大学自动化系,安徽合肥230027 [2]中国科学院合肥智能机械研究所,安徽合肥230031

出　　处：《仪表技术》2013年第9期1-5,共5页Instrumentation Technology

基　　金：国家科技支撑计划课题(2013BAH14F01)

摘　　要：针对健康管理对人体运动方式检测的需求,设计了一种基于加速度曲线形状特征的人体日常体力活动识别系统,识别体力活动包括:休息、上楼、下楼、步行、跑步、乘车。该系统通过采集腰部、外套上口袋、外套下口袋三个位置不同日常体力活动的加速度数据,求取加速度矢量值,滤波降噪后,提取数据分布特征和峰谷值特征描述曲线形状,采用支持向量机建立日常体力活动识别模型。实验表明,所建识别模型受佩戴位置影响较小,可佩戴在腰部、放置外套上口袋、放置外套下口袋三个位置,针对11人数据集使用留一法进行分类测试,平均识别准确率分别可以达到95.96%,95.96%,94.61%;不区分佩戴位置,平均识别准确率也能达到93.60%,有效的对日常体力活动进行了分类判别。This paper present a Recognition system of Daily physical activity based on a 3D acceleration sensor. A novel acceleration curve shape feature was proposed to recognize numerous human daily physical activities, e.g. resting, going upstairs, going downstairs, walking, running, and driving. The sensor signals were obtained from accelerometers on waist, upper and lower pockets on coat. This paper chooses Support Vector Machine to recognize the activities. The experiments show that, effectiveness of the curve shape features of signals, and the average recognition accuracy rate can reach 95.96%, 95.96%, 94.61% at waist, upper pocket and lower pocket, respectively. If we mix the signals from sensors at three positions, the average recognition accuracy rate can still reach 93.60%.

关 键 词：加速度 动作识别 形状 支持向量机 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]