基于无线传感器网络的车辆检测识别算法研究  

Research on vehicle detection and classification algorithm based on wireless sensor networks

作　　者：刘桂林[1] 孔祥维[1] 刘航[1] 

机构地区：[1]大连理工大学电子工程系,辽宁大连116023

出　　处：《传感器与微系统》2010年第2期9-12,共4页Transducer and Microsystem Technologies

摘　　要：针对无线传感器网络(WSNs)的特点,利用车辆发出的声音信号,提出并研究了一种改进的信号检测算法,能够有效地从被噪声严重污染的声音信号中提取出车辆信号。使用小波包变换提取16维信号特征,支持向量机进行目标分类,得到单节点识别结果。提出了基于能量的全局决策融合算法,对多个节点做出的决策进行融合,得到网络的最终识别结果。为了评估算法,使用了来自DARPA SensIT实验中的真实数据,其中包含了履带车和重型卡车的大量声音信号。实验结果表明:该算法用于WSNs中的车辆识别方面是有效的。Aimed at the characteristics of the wireless sensor networks, an improved algorithm is presented for signal dectection in order to reduce the false alarm rate, by using battlefield vehicle acoustic signal. Wavelet packet transform is used to extract 16-dimensional signal feature, support vector machine in classifing target. Energy-based global decision fusion algorithm is proposed to work out the final result. For evaluation purposes, real data are used from DARPA＇ s SensIT project, which contains lots of acoustic signals of two types of vehicle, a tracked vehicle and a heavy truck. The experiment results show that the algorithm is valid for battlefield vehicle classification based on wireless sensor networks.

关 键 词：无线传感器网络 信号检测 车辆识别 决策融合 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]