基于高斯模糊的摔倒自检测算法设计与应用

The design and application of fuzzy based auto fall detection algorithm

摘要:

随着智能设备的普及,如何快速准确地检测、识别人体摔倒已逐渐成为研究的热点。然而现阶段对摔倒动作识别与检测仍然存在很多问题。为此,以智能设备的传感器系统采集的三轴加速度与角加速度为基础,结合经过高斯过滤后形成人体活动的信号幅度向量和陀螺仪信号幅度向量特征曲线与摔倒检测的模糊隶属函数特征模型,提出一种基于模糊的摔倒自检测算法。算法重点针对急速跑动、上下楼梯、手机平抛和自由落体等摔倒检测中的干扰动作进行了分析与区分,经过实验测试表明该算法有较快的反馈速度、较好的区分度以及较低的误判率。

Abstract:

With the popularity of smart devices, how to correctly identify the elderly fall detection, has gradually become the research hot spot. However, at present about falling action recognition and detection still exist many problems. Therefore, intelligent equipment of triaxial acceleration and Angle acceleration of the sensor system acquisition, on the basis of combining forming human activity after gussied filtering signal amplitude vector (SMV) and the gyroscope signal amplitude vector (GSMV) characteristic curve and fuzzy membership function characteristics of the fall detection model, proposing a fall since detection based on fuzzy algorithm. Algorithm focuses on rapid movement, up and down the stairs, mobile phone horizontal cast and a free fall, the interference action of the fall detection are analyzed and distinguish, through experimental tests show that the algorithm has good degree of differentiation and low misjudgment ratio.