便携式人体跌倒检测及自动预警方法

Portable Human Fall Detection and Automatic Warning Method

传统人体跌倒检测和预警方法受地点限制,且采集人体行为数据不可靠,导致检测和预警精度低,实用性差。为此,提出一种新的便携式人体跌倒检测及自动预警方法。通过对人体行走、弯腰、站立、坐下、躺下和跌倒等动作角度进行分析,发现跌倒时人体有显著的角度变化。对采集的角度数据波形阶段进行分割,通过信号时域对波形进行判断。利用K-L变换方法从庞大的特征量集合中获取合理特征量,分别是加速度峰值和倾角变化值,介绍其计算方法,找出最佳判断值。将跌倒检测方法编写成代码在便携式设备上实现,开发便携式设备终端智能APP,实现人体跌倒自动预警。实验结果表明,所提方法实用性强,跌倒检测和预警精度高。

Traditional human fall detection and early warning methods are limited by the location,and the acquisition of human behavior data is unreliable,resulting in low detection and warning accuracy,poor practicability.Therefore,a new portable fall detection and automatic warning method for human body is proposed. By analyzing the angles of walking,bending,standing,sitting,lying down and falling down,it is found that the human body has a significant angle change when falling down. The angle data waveform phase is segmented,and the waveform is judged by the signal in time domain. Through the K-L transform method,the reasonable characteristic quantities are obtained from the huge feature set,which are the acceleration peak value and the dip angle variation value,and the calculation method is introduced to find the best judgment value. The fall detection method is written as code on portable devices,and the portable device terminal intelligent APP is developed to realize automatic fall warning of human body. The experimental results show that the proposed method is practical and has high accuracy in fall detection and warning.