基于声音识别的交通信息检测技术研究

发布时间：2018-02-01 10:28

本文关键词： 车辆噪声交通信息检测声音识别信号处理　出处：《长安大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着我国交通运输业的发展和经济的繁荣，传统的高速公路管理模式已经跟不上高速公路发展的步伐，对交通信息检测的不力，是导致目前交通拥堵、事故频发的重要原因。如何实现交通信息检测的智能化、信息化、高效化，成为目前亟待解决的一个问题。基于声音识别的交通信息检测技术，综合应用数字信号处理技术、机械电子工程技术、计算机技术等，可以在很大程度上弥补电磁感应线圈车辆检测器、电磁波感应检测器和视频车辆检测器等传统车辆检测设备易损坏、破坏路面、受环境影响明显、价格昂贵等不足，大大提高交通信息检测的效果，具有非常重要的现实意义。目前的音频信号识别方法限于技术问题，未达到令人满意的程度，因此，基于声音识别的交通信息检测技术还有很大的研究空间。本课题在查阅了国内外相关文献的基础上，详细分析了车辆声音信号的产生机理及其信号特征，通过与语音信号相对比，分析了利用语音信号处理的方法对车辆音频信号进行处理的可行性。以此为前提，论文对车辆音频信号的预处理及其时频域的特征参数提取做了研究。识别过程中，选用隐马尔可夫模型(HMM)作为识别系统的基础，，阐述了HMM的基本原理及其三个基本问题和解决算法，并提出了基于车辆音频信号进行车型识别的理论方案及检测流程。在车辆声音识别检测系统实验的过程中，选用了驻极体麦克风和AD7606数据采集模块，并采集了东风农用三轮车和大众Sagitar1.4t轿车的通过噪声。用Matlab对采集到的声音信号进行特征提取分析，通过对比两种声音信号的特征，验证了基于声音识别的交通信息检测技术的可行性。
[Abstract]:With the development of China's transportation industry and economic prosperity, the traditional expressway management mode has been unable to keep up with the pace of highway development, the lack of traffic information detection is leading to the current traffic congestion. How to realize the intelligent, information and high efficiency of traffic information detection has become an urgent problem to be solved. The traffic information detection technology based on voice recognition, the comprehensive application of digital signal processing technology, mechanical and electronic engineering technology, computer technology and so on, can make up for the electromagnetic induction coil vehicle detector to a great extent. Traditional vehicle detection equipment, such as electromagnetic wave sensor and video vehicle detector, is easy to damage, destroy the road surface, be obviously affected by the environment, expensive and so on, greatly improve the effect of traffic information detection. The current methods of audio signal recognition are limited to technical problems and do not reach a satisfactory level. Therefore, there is still much room for research on traffic information detection technology based on sound recognition. On the basis of consulting the relevant literature at home and abroad, this paper analyzes in detail the mechanism and characteristics of the vehicle sound signal, and compares it with the speech signal. The feasibility of using the method of speech signal processing to process the vehicle audio signal is analyzed. In this paper, the preprocessing of vehicle audio signal and the extraction of characteristic parameters in time-frequency domain are studied. In the process of recognition, Hidden Markov Model (HMMM) is chosen as the basis of the recognition system. The basic principle of HMM and its three basic problems and solving algorithms are expounded. In the experiment of vehicle voice recognition system, electret microphone and AD7606 data acquisition module are selected. The noise of Dongfeng agricultural tricycle and Volkswagen Sagitar1.4t car was collected and the sound signal was extracted and analyzed by Matlab. The feasibility of traffic information detection technology based on sound recognition is verified by comparing the characteristics of two kinds of sound signals.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U495

【参考文献】