节点文献
无线传感器网络中声震信号的特征提取算法
Feature Extraction Algorithm for Acoustic-seismic Signals in Wireless Sensor Networks
【摘要】 面向无线传感器网络在地面目标声震信号识别方面的应用需求,提出基于局域判别基(Local Discriminant Bases,LDB)算法的特征提取方法.并且,针对现有的基于时频能量图的可分性测度的缺点,提出新的基于概率密度估计的相对微分熵的可分性测度.基于实地采集到的信号的仿真实验表明,该方法在一定程度上提高了目标的正确识别率,降低了特征维数,具有实际的应用价值.
【Abstract】 Feature extraction algorithm based on local discriminant bases (LDB) is proposed to satisfy the requirement of application on classification of acoustic-seismic signals of ground targets in wireless sensor networks. And a new discriminant measure using relative differential entropy which is based on probability density estimation is proposed to solve the problem of time-frequency energy map. Experiments based on real signals indicate that this method can improve the performance of classification at a certain extent and decrease the dimension of features,so it is practically valuable for application.
【Key words】 wireless sensors networks; feature extraction; LDB; time-frequency energy map; relative differential entropy;
- 【文献出处】 小型微型计算机系统 ,Journal of Chinese Computer Systems , 编辑部邮箱 ,2010年02期
- 【分类号】TP212.9;TN912.3
- 【被引频次】1
- 【下载频次】184