基于视频的车流量及车辆闯红灯检测系统研究

发布时间：2018-10-05 11:04

【摘要】：智能交通系统(Intelligent Transportation System)是我国未来交通发展的趋势,车流量统计与车辆闯红灯检测是智能交通系统的重要组成部分,传统的基于地感应线圈的系统需要破坏已经建好的路面,影响道路的使用寿命,而近年来应用越来越广泛的基于虚拟线圈的系统不能对车辆的运动轨迹进行跟踪,因此,本文旨在设计一个基于视频的车流量及车辆闯红灯检测系统,此系统能够准确预测车辆的位置并绘出车辆的运动轨迹,系统有较好的视频处理速度和车辆检测的准确度。本文的主要工作有以下三个方面:1、研究了车辆检测的前期工作,包括图像平滑、背景建模、目标提取及阈值分割几个步骤,并对每个步骤所选用的算法进行研究和仿真对比它们的处理速度和实际效果,最终确定采用中值滤波对图像进行预处理、采用混合高斯法进行背景建模、采用背景差分法进行目标提取以及最大类间反差法进行阈值分割。2、在选定算法后,结合算法所需要的运算量及系统的设计要求,选用三星S5P6818处理器,摄像头选用CMOS,USB接口,并进行了Linux内核移植配置,安装QT和opencv的运行时序。3、利用QT框架和opencv计算机视觉库,对所设计的硬件平台上的配套软件进行开发,用卡尔曼滤波法对车辆位置进行预测并跟踪,设计了系统的交互界面,实现了车流量统计及闯红灯检测的功能。经过实验证明,本系统的准确度和处理速度基本能够达到预期效果。
[Abstract]:Intelligent Transportation system (Intelligent Transportation System) is the trend of traffic development in our country in the future. Traffic flow statistics and vehicle red light detection are important components of its. The traditional system based on earth induction coil needs to destroy the road surface that has been built. It affects the service life of the road, but in recent years, the system based on virtual coil, which is more and more widely used, can not track the moving track of the vehicle. Therefore, this paper aims to design a system for detecting the traffic flow and the vehicle running through the red light based on video. The system can accurately predict the position of the vehicle and draw the moving track of the vehicle. The system has good video processing speed and vehicle detection accuracy. The main work of this paper is as follows: 1. The preliminary work of vehicle detection is studied, including image smoothing, background modeling, target extraction and threshold segmentation. The algorithms selected for each step are studied and simulated to compare their processing speed and actual effect. Finally, the median filter is used to preprocess the image and the mixed Gao Si method is used to model the background. The background difference method is used for target extraction and the maximum inter-class contrast method is used for threshold segmentation. After the algorithm is selected, the Samsung S5P6818 processor and the camera CMOS,USB interface are selected in combination with the computation and system design requirements of the algorithm. The porting configuration of Linux kernel is carried out, the running sequence of QT and opencv is installed. By using QT frame and opencv computer vision library, the matching software on the hardware platform is developed, and the vehicle position is predicted and tracked by Kalman filter method. The interactive interface of the system is designed, and the function of traffic flow statistics and red light detection is realized. Experimental results show that the accuracy and processing speed of the system can achieve the desired results.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP391.41

【相似文献】