-
中文摘要运动车辆的实时检测和跟踪是交通检测的核心问题，基于视频图像的车辆检测跟踪更是研究热点，是光电图像动态目标跟踪的一项重要应用。本文根据交通场景的特点，将静态背景下交通视频图像处理分为车辆识别区域和车辆跟踪区域两部分。车辆识别区中，我们基于虚拟区域视频检测实现了实时的车辆检测，并结合背景差分原理将车辆从背景图像中分割出来。车辆跟踪区中，运动车辆实时跟踪是通过车辆检测，车辆运动估计和车辆匹配三部分实现。引入了投影法估计车辆运动，同时结合Kalman滤波器缩小了匹配时的搜索范围，优化了跟踪算法。另外，本文对车辆检测中出现的典型问题提出了解决方案：根据视频图像中阴影区别于车辆的特点识别并去除阴影；利用多虚拟线检测器解决车辆变更车道等。6649
- 上一篇：SRA硅微谐振式加速度计非线性振动特性研究
- 下一篇：基于圆光栅的莫尔体层析投影特性研究
关键词 交通检测系统，车辆检测，车辆跟踪，视频图像处理
毕业设计说明书（论文）外文摘要
Title Detection of moving target in optoelectronic image sequences
Abstract
Detection and tracking of moving vehicles based on video detection is by
far one of the research hotspot in vision vehicle detection and an
application of optoelectronic moving image target detection. Considering
the features of traffic scene in static background, we pide the
processing of video sequences into two parts: vehicle detection area and
vehicle tracking area. In the vehicle detection area, an algorithm based
on virtual areas is adopted in real time vehicle detecting. We also segment
target vehicles from a frame of video sequences using the background
differencing method. In the vehicle tracking area, the real time tracking
of moving vehicle is realized by vehicle detection, motion estimation and
vehicle matching. We introduce a projection method and present an algorithm
of vehicle tracking based on Kalman filter, which helps us achieve great
algorithm optimization through decreasing the size of hunting zone in
vehicle matching part. Besides, we provide solutions to some typical
problems in vehicle detection: We investigate into the differences between
shadows and vehicles in video sequences to recognize and remove shadow;
We develop a detector based on multiple virtual lines to deal with
situations like changing of vehicle lane and so on.
Keywords Traffic detecting system, vehicle detection, vehicle tracking,
video image sequence processing
目 次
1 绪论 1
1.1 研究背景及意义 1
1.2 国内外研究现状 1
1.3 课题研究内容及方法 2
2 图像处理方法概述 4
2.1 彩色图像灰度化及去噪 4
2.2 图像纹理和边缘检测 5
2.3 灰度图像的二值化 6
2.4 二值化图像的形态滤波 8
2.5 连通域分析 10
3 运动车辆检测和跟踪实现 11
3.1 背景的提取和更新 11
3.2 基于虚拟线检测车辆 14
3.3 运动车辆阴影检测 15
3.4 计数检测器 16
3.5 基于卡尔曼预测模型的运动车辆检测 17
3.6 运动车辆检测和跟踪流程 22