针对老龄人异常行为的实时检测、识别和在线主动信息推送问题，利用智能移动终端内置的三轴加速传感器来采集人体运动信息。通过匹配追踪算法（MP）对信号进行Gabor原子分解，并利用Wigner-Ville时频分析方法，从时间和三维空间动作特征对其进行时频联合域研究。为解决时频分析过程中的复杂运算问题，并完成异常行为动作的在线训练，分类识别和信息推送，探讨一种运用移动终端APP，无线网络和云计算平台来构建的可穿戴式老龄人异常行为检测系统。实验结果表明，该方法切实可行，可将其应用于老龄人日常监护和紧急救助等相关领域。%Aiming at the problems of the abnormal behavior real-time detection,recognition and online push service for the elderly. Intelligent mobile terminals with tri-axial acceleration sensor embedded are used to capture the motion information of human. The data are decomposed into a linear combination of Gabor atoms by the method of matching pursuit. The Wigner-Ville time-frequency analysis method is introduced and the problem is studied by joint time-frequency analysis in a real time space and by the motion characteristics in 3D space. In order to solve problems of complicated computing in the process of time frequency analysis,and conducting data training,classification,recognition and message push online of abnormal behavior,a wearable abnormal behavior detection system is explored based on the mobile terminal APP,wireless network,and the cloud computing platform. Experimental results demonstrate that the proposed method is feasible and it can be applied to the aged care,emergency aid and related fields.
展开▼
