摘 要
计算机视觉领域中的运动目标检测与跟踪,是从视频或图像序列数据中检测出
运动对象并进行持续追踪,它为高层的视频理解提供基础元素和分析依据。本文研
究并实现了视频中利用帧间差分算法和背景差算法进行运动目标检测,分析了各种
检测方法的优缺点。利用Kalman滤波器的原理,结合波门跟踪,对所采集的视频进
行了单个目标跟踪实验,跟踪效果良好;针对同一个视频中多个运动目标的跟踪问
题,本文利用SIFT算法对感兴趣运动目标进行锁定跟踪实验,结果表明,当有多个
运动目标出现时,对感兴趣的目标均有良好的跟踪效果。
关键词: 目标检测,目标跟踪,Kalman 滤波,SIFT 算法
ABSTRACT
The task of moving object detecting and tracking in the computer vision system is to
detect the moving target in the in video sequence or image sequence and keep tracking. It
provides basic elements and analysis foundation for the high-level understanding of videos.In
this paper, the detecting of moving target was reseached and achieved using the methods of
inter-frame difference and background difference. It also analyzed the advantages and
disadvantages of various detection methods. The tracking of moving target was accomplished
by the method using Kalman filter which combined the gate-tracking principle, and got a
good result in the experiment from the collected video with a single target . An algorithm
based on SIFT algorithm for moving target tracking is proposed in this paper. The algorithm
sovled the problem in the tracking target when multiple moving objects appreared in one
video. The result showed that tracking one target in the video scene where multiple moving
targets appeared was achieved.
Xie Hao(Communication and Information System)
Directed by associate prof. Yu Ping
KEY WORDS: moving target detecting, moving target tracing, Kalman filter, SIFT
algorithm
目 录
中文摘要
英文摘要
第一章 绪论 1
1.1 课