基于SVM和阈值分析法的摔倒检测系统  

FALL DETECTION SYSTEM BASED ON SVM AND THRESHOLD ANALYSIS

作　　者：陈玮[1] 周晴 曹桂涛[3] 

机构地区：[1]华东师范大学教育信息技术学系,上海200062 [2]上海浦东发展银行上海分行,上海200000 [3]华东师范大学计算机科学与软件工程学院,上海200062

出　　处：《计算机应用与软件》2017年第7期182-187,276,共7页Computer Applications and Software

基　　金：国家重点基础研究发展计划项目(2011CB707104)

摘　　要：随着我国人口老龄化的快速发展,老年人口呈现出高龄化、空巢化的趋势。当老年人在家中发生意外跌倒而未能及时获得救助时,会给老年人造成严重的身心伤害。针对这个问题,设计并实现老年人摔倒检测系统。该系统以嵌入式微处理器K60核心开发板作为处理内核,加速度传感器MMA7660FC采集人体三轴加速度信息,ENC-03陀螺仪采集两轴角速度信息。通过基于支持向量机(SVM)和阈值分析法的摔倒检测算法判断是否摔倒,在摔倒时能自动地发送摔倒报警信息。实验结果表明,系统能有效地区分摔倒和其他日常生活行为,算法准确度高、实时性高。With the rapid development of aging population, China＇s population gradually shows the trend of aging and empty nest phenomenon. Once if the aged fall to the ground, they haven＇t been discovered soon enough and take reasonable measures immediately; it would bring serious physical and psychological harm to them. In order to solve this problem, we design and implement the fall detection system for the elderly. The system uses the embedded microprocessor K60 core development board as the processing core, the accelerometer MMA7660FC collects the three- axis acceleration information of human body, and the ENC-03 gyroscope gathers the angular velocity information of the two axes. A fall detection algorithm based on SVM and threshold analysis is used to judge whether the old man falls down or not, and it can automatically send the falling alarm information when falling. Experimental results show that the system ＇can effectively distinguish between falls and other daily life behaviour. The algorithm has high accuracy and high real-time performance.

关 键 词：加速度传感器 陀螺仪 支持向量机 阈值分析法 摔倒检测 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]