车辆个数估计与车型分类的声频算法研究

【作者】 罗向龙；

【导师】 郭元术；

【作者基本信息】 长安大学 ， 交通信息工程及控制， 2003， 硕士

【摘要】 本文从分析现行交通量检测和车型识别的现状出发，提出了一种利用车辆声音信号进行交通量检测和车型分类的新方法。对麦克风阵列所采集的车辆行驶时的声音信号，运用宽带阵列信号处理的方法进行离线处理，并将车辆个数的估计转化为阵列信号处理中信号源个数的估计，将车辆上下行的估计转化为阵列信号处理中波达方向变化的估计，进而较准确地估计出了车辆的个数与行车方向；同时对所采集的车辆声音信号应用模式识别中的主分量分析法实现了车辆的简单分类，为实现车型识别作了一些初步的探索。最后通过实验论证了上述方法的可行性。更多还原

【Abstract】 In this dissertation, according to the condition of current traffic volume detection and vehicle recognition, a new method of traffic volume detection and vehicle classification with vehicle noise is presented. The noise signal of passing vehicles is gathered with microphone array, and processed off line with the method of wide-band array signal processing, then the estimation of vehicle numbers can be transformed into the estimation of signal source numbers, up and down of the vehicle can be transformed into DOA estimation of array signal processing. At the same time, vehicle types are classified using PCA method of the pattern recognition based on vehicle noise, and some primarily study is done for vehicle recognition. At last, the result of the experiment proves this method is feasible.更多还原