摘要：随着人民生活水平的不断提高，随着世界经济的发展，人们对车辆的需求只增不减，然而车辆增加带来的安全隐患也是不容忽视的，如何有效地监控是解决安全问题的关键点之一。在这种背景下，我们就需要车辆检测系统来实时地监控车辆。车辆检测系统主要是基于视频检测技术和运动物体检测算法，前者负责解决难题并加强效果，后者负责提取运动物体。本课题的主要研究内容是车辆检测系统的三种传统算法：帧差法，光流法，背景差分法的检测原理，给出它们的算法流程，利用C++结合OpenCV分别实现这三个功能，针对视频进行实验。得到实验结果后需要对实验结果进行具体分析，讨论得出它们各自的优缺点，再和其公认的优缺点进行对比，看看是否一致。43489

毕业论文关键词：帧差法；光流法；背景差分法；车辆检测

Comparative study on the effect of the traditional algorithm in vehicle detection system

Abstract: With the improvement of people's living standards, with the development of the world economy, people come to be more and more need vehicles, but the security risks with vehicle’s increase can not be ignored, how to effectively monitoring the vehicles is one of the the way to solve the security problem. In this context, we need a vehicle detection system to monitor the vehicles in real time. Vehicle detection system is mainly based on video detection technology and moving object detection algorithm, the former is responsible for solving problems and strengthen the effect, the latter is responsible for extracting moving objects. The main research contents of this paper are three traditional algorithms of vehicle detection system: Frame Different Method, Optical Flow method and Background Subtraction, their detecting principle. I will use C++ combined with OpenCV to achieve these three functions, and shoot a video to experiment. After getting the results, we need to not only make a concrete analysis for the results, but also discuss their respective advantages and disadvantages, and then compare with the industry idea to see whether it is consistent with the results.

Keywords: Frame Different Method; Optical Flow; Background Subtraction; Vehicle detection

目录

摘要

Abstract i

目录 ii

1 绪论 1

1.1 课题的背景与意义 1

1.2 国内外研究现状与发展 2

1.3 本文主要研究内容 3

1.3.1 课题基本内容 3

1.3.2 课题重点 4

1.3.3 课题难点 4

1.3.4 本文工作 4

2 图像处理技术 6

2.1 图像介绍 6

2.2 图像处理 6

2.2.1 图像压缩 6

2.2.2 图像增强和复原 7

2.2.3 数学形态学 10

3 传统运动物体检测算法 12

3.1 传统运动物体检测方法概述 12

3.2 光流法 12

3.2.1 光流法的检测原理 12

3.2.2 光流法分类 13

3.2.3 光流法的限制条件