摘要
根据人体在热释电红外(PIR)探测器的检测区域内沿不同路径和不同方向运动时信号在时域及频域的分布特点,提出一种基于单只PIR探测器信号的人体运动特征识别方法。首先提取人体PIR信号的频谱和短时频谱能量特征;然后进行主元分析(PCA)特征降维,根据典型相关分析(CCA)进行特征融合;最后采用最小二乘支持向量机(LS-SVM)方法进行分类识别。实验以不同人体、不同运动方式的PIR探测器数据为研究对象。分析结果表明,提出的特征提取、特征融合及识别方法能有效地对人体运动特征进行识别。
Based on the distributions of a human walking within the view field of a pyroelectric infrared(PIR) detector along different paths or directions,a human motion recognition method using output signal of single PIR detector is presented.Fourier transform and short time Fourier transform(STFT) energy of the PIR signal are calculated as motion features,then principle components analysis(PCA) is used to reduce the feature dimensions,and canonical correlation analysis(CCA) is applied to feature fusion.Finally,least square support vector machine(LS-SVM) is adopted to classify the feature vectors.Experimental results show that the proposed feature extraction,feature fusion and recognition method are efficiently for human motion recognition.
出处
《光电子.激光》
EI
CAS
CSCD
北大核心
2010年第3期440-443,共4页
Journal of Optoelectronics·Laser
基金
国家"863"计划资助项目(2007AA01Z423)
国家"十一五"基础研究资助项目(C10020060355)
重庆市科技攻关计划资助项目(CSTC2007AC2018)
重庆市重点科技攻关项目(CSTC2009AB0175)