摘要
针对热释电红外(PIR)传感器在室内人体定位及识别上的准确率问题,设计了一种人体红外信号感知模型,提出了一种定位与识别的新型方法。模型节点采用一对正交的PIR传感器,结合对菲涅尔透镜的视场角调制,能够有效探测水平与垂直方向的人体红外信号。通过对这一对PIR传感器时域输出信号的采集分析,采用时域信号的峰值时间序列特征并融合两只传感器数据的相关性分析,使用机器学习SoftMax分类方法进行位置及速度等级的分类。实验结果表明:所设计方法在位置与速度等级分类上实现94.79%的准确率,在室内场景智能感知上具有较好的应用价值。
Aiming at the accuracy of pyroelectric infrared(PIR)sensor in indoor human body positioning and recognition,a human body infrared signal sensing model is designed,and a new method of positioning and recognition is proposed.This model node uses a pair of orthogonal PIR sensors and combines with the field angle modulation of the Fresnel lens,to effectively detect human infrared signal in the horizontal and vertical directions.This method through acquiring and analyzing on time-domain output signals of this pair of PIR sensors,using peak time series characteristics of time-domain signals and correlation analysis of fusion of two sensor data,and using SoftMax classification method of machine learning algorithm to classify positions and speed levels.Experimental results show that the design method achieves accuracy of 94.79%in levelclassification of position and speed,and has good application value in intelligent perception of indoor scenes.
作者
徐晓冰
孙百顺
孙伟
左涛涛
焦宇浩
XU Xiaobing;SUN Baishun;SUN Wei;ZUO Taotao;JIAO Yuhao(School of Electrical Engineering and Automation,Hefei University of Technology,Hefei 230009,China)
出处
《传感器与微系统》
CSCD
北大核心
2021年第8期38-41,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(51877060)。
