学位论文---基于背景差分法的物体识别.docVIP

基于背景差分法的物体识别 OBJECT RECOGNITION BY BACKGROUND DIFFERENCE METHOD 专 业：　自动化 摘 要 图像识别技术在很多领域都是产业现代化的基础，所以在当今社会它成为了研究的热门学科。由于被测物体会由于时间的不同，背景的不同，气候的不同,被测物体会表现出不同的外形特征，具有错综复杂和瞬间转变的性质［1］，所以图像识别的准确率一直是一个难以解决的课题。而基于背景差分的图像识别技术则能很好地解决这一困难的问题，它是整个图像识别领域中最基础并且可靠性最高的一门科学［2］。这种方法以将背景图片与实时图像作差分的方法有效提高普通图像识别方法的准确度［3］，也可以对运动物体做出准确性较高的识别，并且具有较快的处理速度，能在OpenCV1.0的帮助下完成目标。目前，基于背景差分的图像识别主要应用于车辆流量检测技术［4］、车牌检测［5］、弱信号的目标检测、稻米粒优劣的判断、运动人体检测等领域。本文使用USB外接摄像头从外界获得背景与被测视频流，在Windows系统下利用Visual Studio C++6.0与OpenCV 1.0结合编程的方法［6］，对背景与被测视频流做差分，从而达到物体识别的目的。由于使用的摄像头分辨率较低，所以测量的结果精确度并不是很高，测量物体也较为简单，如何利用高分辨率的摄像头以及高版本OpenCV函数来识别更复杂的被测物体是今后的研究目标。 关键词： 背景差分； 图像识别； Opencv；Visual Studio C++6.0 ABSTRACT Image Recognition has been one of the most popular deciplines because that it’s the key of Industrial modernization in numerous fields. The measured objects will be in difference shape for the difference of seasons; background and temperature.They are complex and variable, thus, it’s really a tough work to increase the accuracy of Image Recognition. Image recognition technology based on background subtraction, a subject that is very reliable and accurate in the Image Recognition field, solves that problem well.This method improves the accuracy effectively by differencing the video stream to a background picture, it also can identificat the moving subject effectively, it can progress a subject in high speed by the using of the opencv 1.0. At present, Image Recognition is often utilized in the field such as vehicle traffic detection technology，target detection of weak signals, judge the merits of rice and Human Movement detection,there we get backgroud and video stream from the outside world by USB external camera, programme with opencv 1.0 by visual studio C++6.0 in the operating system of Windows.Then we identificate a subject by differencing backgroud and video stream.The acurracy of Image Recognition is below the ideal situation because of the resolution of USB