Meanshift基于OpenCV的夜间运动目标跟踪方法研究

菜单

摘要计算机视觉和数字图像处理技术在军用民用、工程、航空和监控上面得到广泛的应用，其中研究前景最有前途的一个方面是连续视频图像中对物体进行分析，在机器人识别方位、智能视觉监控系统、电学图像分析、农工业检测、视频图像分析以及军用设备视频信号的处理上都有着主导地位。

其中特别复杂的一个地方是基于夜间运动目标跟踪分析，这个同时也是计算机视觉和数字图像处理技术的难点。它处理量十分浩瀚，不仅在采集信号的同时，又要处理已经采集的图像，还要进行开发针对应的应用编写程序。对那些工程人员来说，如果要做一些项目，这些项目的基础算法都需要自己编写，重新编写全部图像处理函数，不仅浪费时间精力，而且稳定性难以得到保证。

本文了解夜间运动目标特性，掌握夜间运动目标跟踪的基本原理，选取视频运动分析中的运动目标跟踪作为研究重点。利用OpenCV和Visual C++组建实验平台，针对数字图像特定噪声的去除。运动物体的正确识别和分割、物体之间的遮挡、处理的实时性要求等难点进行了研究。研究Meanshift法、粒子滤波法以及光流跟踪法三种目标检测的算法，利用C语言，并结合OpenCV函数库，编程实现三种跟踪算法，调试程序，分析算法结果并比较各种算法的优缺点及其适用范围。65166

毕业论文关键词：视频运动目标检测 Meanshift 粒子滤波光流跟踪 OpenCV

毕业设计说明书（论文）外文摘要

Title OpenCV-based the night moving target tracking Method research

Abstract

Computer vision and digital image processing technology can be widely used in industry, healthcare, aerospace, military and other fields, including continuous motion object in the image for video analysis is one of the most widely application prospects, in robot navigation, intelligent visual monitoring system, military radar video image analysis and video signal processing occupies an important position.

The movement target tracking based on night analysis and computer vision and digital image processing technology is the most complicated one direction. It is a big project, both for signal acquisition, and processing to the acquisition of images, but also for specific application program development. For engineering personnel to complete an application, both caused the waste of time and energy, and to guarantee stability, practicality and generality.

This understanding of the night moving target characteristics, master nighttime basic principle of moving target tracking, select the video motion analysis of moving target tracking as a research priority. Using OpenCV and Visual C + + build experimental platform specific for digital image noise removal. Correctly identify moving objects and segmentation, between objects obscured the real-time processing requirements difficulties were studied. Meanshift research method, particle filter and optical flow tracking method three kinds of target detection algorithm, using C language, combined with OpenCV library, programming, achieve three tracking algorithm, debugger, and compare the results of analysis algorithm advantages and disadvantages of various algorithms its scope of application.

Keywords: Moving target Target tracking OpenCV meanshift Particle filter method Tracking of optical flow method

1 绪论 1

1.1本文研究背景及意义 1

1.2 国内外研究现状