基于运动检测的家用智能监控系统的研发

电气自动化论文编号:ZD1066    论文字数:24610,页数:45

目 录
摘要……………………………………………………………………………………Ⅰ
一 绪论……………………………………………………………………………………………1
   1.1 选题背景…………………………………………………………………………1
   1.2视频监控的概述……………………………………………………………2
   1.3论文主要工作及结构…………………………………………………………5
二 基于Opencv的图像采集与处理………………………………………………………6
   2.1 Opencv简介…………………………………………………………………6
     2.1.1 Opencv的特点…………………………………………………………………6
     2.1.2 Opencv的新功能…………………………………………………………7
   2.2 图像处理的技术………………………………………………………………7
   2.3 图像处理的基本要求……………………………………………………8
   2.4 图像处理的举例……………………………………………………………9
     2.4.1 图像数据读取…………………………………………………………………9
     2.4.2 图像灰度化…………………………………………………………10
     2.4.3 图像的中值滤波………………………………………………………11
     2.4.4 图像二值化………………………………………………………12
     2.4.5 图像细化………………………………………………………………13
   2.4图像处理的应用………………………………………………………14
三 运动检测的设计与开发…………………………………………………15
   3.1 功能特点与要求…………………………………………………………15
   3.2 界面功能使用流程图………………………………………………………17
   3.3 运动检测方法…………………………………………………………………17
     3.3.1光流法…………………………………………………………………18
     3.3.2帧差法………………………………………………………………20
     3.1.3 背景消减法………………………………………………………………21
   3.4 背景减除和帧差结合检测目标……………………………………………23
   3.5 系统设计实现……………………………………………………25
     3.5.1 设计目标……………………………………………………………25
    3.5.2 系统的功能…………………………………………………………25
    3.5.3 系统的功能的实现……………………………………………………25
         3.6 智能监控试验结果…………………………………………………………28
 四 总 结…………………………………………………………………………………31
致 谢……………………………………………………………………………………………32
参考文献……………………………………………………………………………………33
附录……………………………………………………………………………………………34

摘  要
 随着科技的发展，家庭监控系统逐渐成为家庭安防的重要手段。家庭监控系统的提高越来越受社会的瞩目。本文设计了一套基于运动检测的家庭智能监控系统，设计主要结合家庭监控几个特点：范围小、精度高、可远程监控、制作成本低，适用于大众家庭。
 在家庭监控方面运动目标检测多是基于底层视频信息的检测，由于大多数图像处理过程中考虑的是对应运动区域的像素，所以运动区域的有效提取对于目标分类、目标跟踪以及行为理解等工作具有重要的意义。运动目标检测有极其广泛的应用范围，比如:军事侦察、智能监控和交通检测等领域。用于运动目标检测的序列图像可分为静止背景、运动背景两种情况，本文是基于运动背景的研究。
 文中首先概要介绍了运动目标检测的几种常用方法，包括光流法、帧间差分法、背景消减法，对各种方法进行了比较，并分析了其优缺点。在此基础上本文提出一种静止背景中基于边界的运动目标检测算法，本算法首先计算相邻两帧图像差分图像的闭值，将差分图像转化为二值图像，然后对二值图像进行形态学滤波确定运动目标的粗略位置，同时对当前帧进行边缘检测，将此边缘检测结果与前者确定的运动目标粗略位置进行交运算，得到运动目标的边界，再经滤波、区域填充等处理，最终获得运动目标的区域。

关键词：图像处理、运动检测

Abstract
 With the development of science and technology, home monitoring security system has gradually become an important means for the family. Home monitoring system to improve more and more attention by the community. This article has designed a family-based intelligent motion detection surveillance system designed to combine home monitoring several characteristics: small, high precision, remote monitoring, the production of low cost, applicable to public family.
 Monitoring in the home moving target detection is based on the bottom of the video and more information on testing, as the majority of image processing is taken into account the corresponding movement of the pixel region, the regional campaign for the effective extraction of the target classification, target tracking as well as the work of understanding importance. Moving target detection has extremely broad range of applications, such as: military reconnaissance, intelligence surveillance and testing in areas such as transport. For moving target detection in image sequences can be divided into a static background, the background of both cases movement, this article is based on the movement of the background research.
 First, an overview of moving target detection of several commonly used methods, including optical flow method, inter-frame difference method, the background reduction method, on the various methods were compared and analyzed the advantages and disadvantages. On this basis this paper, a static background of the moving target based on boundary detection algorithm, the algorithm for computing the first two images adjacent to the closure of the value of difference image, the difference image into a binary image, and then to the binary image to form filter learning to identify the rough location of moving targets, while the current frame edge detection, edge detection results to determine the movement of the former position to pay target rough calculation, to be moving targets in the border, and then by the filtering, filling, such as regional processing, the final moving target access area.
Keywords：Image acquisition, motion detection