基于时态差分方法的视频图像中运动目标的检测毕业论文

 2021-04-15 12:04

摘 要

对运动目标的检测是视觉研究领域的一个热点,俨然成为一个重要课题,它是图像处理和计算机视觉学科的重要分支,也是智能监控系统的核心部分,并且广泛应用于机器人视觉、军事制导、视频监控、智能交通等领域中。

本文采用简单的帧间差分法来检测运动目标。首先,依据帧差法检测运动目标的原理。然后基于其原理,以帧差法的程序实现为目的,给出实现步骤和实现流程。最后,基于OpenCV在Visual C 开发平台上开发了帧差法检测运动目标的软件程序。采用此软件对视频图像进行试验,试验结果表明,帧差法可以简单检测出视频图像中的运动目标,但是目标中存在空洞,不完整,而且准确性不高。

针对帧差法检测运动目标的准确性讨论,本文采用了对称帧差法。首先给出了对称差分的原理。其次对对称差分方法进行程序实现,给出程序实现的步骤和流程图。在Visual C 开发平台上基于OpenCV开发了对称差法的软件程序。最后采用不同的视频图像进行试验,试验结果表明,对称差法可提高检测的准确性,但依然存在有空洞,不能检测出完整的运动目标。

关键词:视频图像;运动目标;帧间差分;对称帧差

Moving Object Detection Based On Tense Difference

Abstract

The moving target detection is research focus of the vision field. It has become an important subject and one of the most important branches in image processing and computer vision. At the same time, it is a key part of intelligent surveillance system and is widely applied in many fields, such as robot vision, military guidance, visual frequency monitoring, intelligent transportation , so on.

A simple frame difference algorithm is employed to detect moving objects in this thesis. Firstly, the principle of frame difference algorithm is proposed. Then, the implementation of the frame difference algorithm is given, and the implementation steps and implementation flow chart are given. Finally, the software for detecting moving targets is compiled on Visual C platform based on OpenCV to realize the frame difference algorithm. This software is applied to testify the algorithm with different videos. Experiment results show that the frame difference algorithm can detect the moving objects in simple video images, but the detected targets with holes are incomplete usually, and the accuracy is very low.

Because the frame difference algorithm can not detect moving objects accurately in some case, the symmetrical frame difference algorithm is employed in this thesis. Firstly, the principle of symmetric difference is given. Secondly, the symmetric difference algorithm is implemented, and the steps and flow chart of the program are given. On the Visual C platform, the software program of the symmetric difference algorithm is developed based on OpenCV. Finally, different video images are used to testify the algorithm. The results show that the algorithm can improve the accuracy of detection, but it still has holes and can’t detect the complete moving objects.

Key words:Video Image; Movement Object; Frame Difference; Symmetrical Difference

目 录

1 绪论 1

1.1 研究背景 1

1.2 国内外研究现状 2

1.2.1 光流法 2

1.2.2 背景差分法 3

1.3 研究内容安排 4

2 预备知识 6

2.1 颜色空间 6

2.1.1 RGB颜色空间 6

2.1.2 HSV模型 6

2.1.3 彩色图像到灰度图像的转换 7

2.2 图像滤波 9

2.2.1中值滤波 9

2.2.2 均值滤波 10

2.3 形态学处理 10

2.4 二值化 12

3 帧差检测算法的设计和实现 13

3.1 帧差法原理及优缺点 13

3.2 帧差法步骤和实现 15

4 对称帧差检测算法的设计和实现 20

4.1 对称帧差的原理及优缺点 20

4.2 对称帧差步骤和实现 22

5 结论和展望 28

致 谢 30

参考文献 31

1 绪论

1.1 研究背景

视觉是人类拥有的最基本、最直接观察外界环境的感官,可以应用于人类生活的各个方面。所以近些年来,随着计算机视频处理技术的不断发展,计算机视觉这项研究也得到了许多国内外学者的关注。其中一个重要课题就是运动目标检测与跟踪技术,它包含了人工智能、图像处理、计算机、自动控制、模式识别[1]等诸多方面的前沿技术。在这些方面,运动目标检测与跟踪技术是计算机视觉中比较重要的一个方面,在智能交通、视频监控、智能交通、工业产品检测、视觉导航、医疗诊断、军事制导[2]等实际场景中应用偏多。针对运动目标检测与跟踪技术进行研究,既对创建齐全的视频目标跟踪体制有帮助,同时在某些方面也可以帮助其他相关学科的进步。所以,开展这方面技术的研究有相当高的学术价值。

(1)军事领域

您需要先支付 80元 才能查看全部内容!立即支付

课题毕业论文、开题报告、任务书、外文翻译、程序设计、图纸设计等资料可联系客服协助查找,优先添加企业微信。