论文部分内容阅读
本文基于MMA7260QT加速度传感器获取的人体运动加速度信号,采用人体加速度向量幅值(SVM)和人体加速度向量区域值(SMA)描述了老年人的运动状态,检测人体跌倒,具有良好的准确性和实时性。采用bior3.3小波分析,在轮廓的基础上,最大程度上保留了细节,有效的去除噪声对特征量的干扰。本文提出了人体跌倒检测算法,大大降低了误判率和漏判率。首先,检测人体SVM是否超过阈值进行第一级跌倒检测,区别出人体日常活动(ADL)和跌倒;其次在此基础上,检测第一级各个跌倒的SMA值,是否超过阈值,判断跌倒和疑似跌倒。当两次判断都检测到跌倒发生时,报警。
Based on the human body motion acceleration signal acquired by the MMA7260QT accelerometer, this paper describes the motion state of the elderly by using the body acceleration vector magnitude (SVM) and the human body acceleration vector region value (SMA) to detect human fall, with good accuracy and real-time . Using bior3.3 wavelet analysis, on the basis of the contour, the detail is retained to the maximum extent, and the interference of the noise on the feature amount is effectively removed. In this paper, the human fall detection algorithm is proposed, which greatly reduces the false positive rate and the false negative rate. First, the detection of human SVM exceeds the threshold for first-level fall detection, the distinction between human activities (ADL) and fall; secondly, on this basis, the detection of the first fall of each SMA value exceeds the threshold to determine the fall and suspected Fall. When the two judgments have detected the fall occurred, the alarm.