基于视频的车辆检测理论与方法分析-计算机应用技术专业论文.docx
文本预览下载声明
摘要
摘
要
I
I
随着计算机数字图像处理技术的发展,视频图像处理技术也广泛的应用到车辆检测
中。视频车辆检测负责采集处理道路车辆信息,是智能交通系统的信息服务系统,为交 通控制管理提供数据依据。本文以静止摄像头拍摄的交通视频为研究对象,利用计算机 图像处理方法,提取视频中的运动车辆数据。
本文根据视频车辆检测系统各步骤功能,分别设计了相应的数据结构和算法。车辆 检测与跟踪主要分为背景提取、运动区域提取、车辆位置尺寸提取和车辆跟踪。首先对 图像进行灰度化和去噪等预处理,在背景提取步骤,对 3 类背景提取算法进行比较后, 选用中值法进行背景提取,利用背景图像中的车辆余迹,提出了检测区间提取算法,为 区间背景差分做准备。在运动区域提取步骤,提出了基于检测区间的背景差分法,并利 用差分结果进行了检测区间的修正;在分割阈值选取时,采用了基于像素频数分布的最 佳阈值选取算法;在阴影去除时,提出了基于阴影像素分布特征的单边侧的阴影检测算 法和阴影去除算法;在连通区域提取时,采用轮廓提取算法,并对其进行改进,使其适 用于多目标提取,有效地提取多连通域的外接轮廓。在车辆位置尺寸提取步骤,提出了 矩形整合算法,大大提高了车辆外接矩形和视频车辆的匹配程度。在车辆跟踪步骤,提 出了链表嵌套的数据结构和新的相似系数计算方法,利用矩形中心距离和面积开方的比 值作为相似系数,有效地去除了目标位置对目标外接矩形尺寸的影响,从而能够对运动 车辆进行有效地跟踪。
本文对提出的算法在 VC++开发环境下进行了实验,比较了改进算法和原有算法的 性能,结果表明,本文提出的新算法行之有效,而且比原有算法性能上有了很大提高。
关键词:车辆检测,区间背景差分,阴影去除,车辆跟踪
Abstract
Abstract
Comparing
Comparing the performance of original algorithm with improved algorithm in the VC++
development environment, experiments demonstrate that the new algorithm is extraordinarily efficient, and improved greatly in performance.
Key words:Vehicle detection; background subtraction; Shadow Elimination; vehicle
tracking
III
III
IV
IV
With the development of computer digital image processing, video image processing technology has been widely applied in traffic detection. Video vehicle detection which is responsible for collecting and processing traffic information, is the information services system of Intelligent Transportation System, and provide data support for control and management of traffic. In this paper, we achieve the data of moving vehicle based on computer image processing method, by using the traffic video recorded through a stationary camera.
In this paper, according to the function of each step in video vehicle detection system, the corresponding data structure and algorithm are respectively designed. Vehicle detection and tracking consists of these steps: background extraction, motion region extraction, vehicle position and size extraction and vehicle tracking. After the comparison of 3 kinds of background extracti
显示全部