面向车流量检测的视频图像处理算法的研究.pdf

摘要 随着交通系统的发展，由交通所引起的安全、拥挤、污染、能耗等问题日益严重，如 何提高交通管理水平和交通系统运行效率是近十年来交通行业关注的焦点。车流量检测作 为发展交通科技的关键技术，是目前智能交通系统研究的热门课题。本文围绕融合帧差法 和背景差法对车流量检测系统展开了深入的研究。 本文首先对图像的灰度化、图像的均衡化、灰度图像的增强和图像滤波等数字图像处 理的基本理论进行了介绍。分析了车流量检测的常用算法，在详细地比较了帧差法和背景 差法各自不足的基础上；提出了基于帧差法和背景差法的图像处理融合算法，该算法可以 融合两者的优点，避免各自的缺点，准确率更高。在分析了多帧平均法、分块法等常用方 法适应性的基础上，采用统计中值法对背景进行建模，实现背景估计与背景更新。论文最 后提出了一个基于视频图像处理技术的车流量检测系统的实现方案，并编制了车流量检测 的图像处理软件。 该系统设计是在Visual C++的编译环境下实现的，对采集到的视频数据用车流量检测 算法进行实验。实验与分析发现，在图像的质量较高、虚拟线圈分布合理、阈值选取合适 等条件下，算法的识别率较高，取得了比较满意的效果。 关键词：车流量; 帧差法; 背景差法; 虚拟线圈 I Abstract Along with the development of transportation system, the problems caused by traffic safety, congestion, pollution, energy consumption are getting more and more serious. How to improve the level of traffic management and transportation system operation efficiency is in the limelight for the transportation industry over the past decade. As the key technique for the development of transportation technology,traffic flow detection is a hot research topic of intelligent transportation system. This paper focuses on fusion of frame differential method and background subtraction method to develop in-depth research of traffic flow detection system. Firstly，This paper introduced the basic theory of digital image processing: the gray change of images, the balance processing of image, enhancement of gray image and filtering of image. Afterwards, analysis common algorithms of vehicle traffic detection. Comparing with shortcomings of the frame differential method and background subtraction method particularly, it proposes image processing fusion algorithm which based on frame differential method and background subtraction method. The algorithm can blend the advantages of both, and avoid their res