信息工程毕业论文

江西财经大学自学考试毕业论文题目图像边缘检测方法研究专业信息工程学生姓名肖鹏准考证号036910100101指导教师李开萍二零一一年七月东华理工大学长江学院毕业设计（论文）摘要摘要数字图像边缘检测是图像分割、目标区域识别和区域形状提取等图像分析领域十分重要的基础，是图像识别中提取图像特征的一个重要方法。边缘中包含图像物体有价值的边界信息，这些信息可以用于图像理解和分析，并且通过边缘检测可以极大地降低后续图像分析和处理的数据量。图像理解和分析的第一步往往就是边缘检测，目前它已成为机器视觉研究领域最活跃的课题之一，在工程应用中占有十分重要的地位。经典的边缘检测方法如：Roberts，Sobel，Prewitt，Kirsch，Laplaee等方法，基本上都是对原始图像中像素的小邻域构造边缘检测算子，进行一阶微分或二阶微分运算，求得梯度最大值或二阶导数的过零点，最后选取适当的闭值提取边界。但这些算法均存在对噪声敏感、不能自适应选择闭值、检测效果不太理想等缺点。本文对边缘检测理论和算法作了深入的研究，在具体分析各类传统的边缘检测算法的基础上，重点研究了Canny算法，并结合改进的MTM算法及Otsu算法对Canny算法中的滤波方法和双门限选取方法进行改进。最后，用MATLAB7.0实现该算法，实验结果表明，改进后的算法(CMO算法)取得比传统的Canny算法更好的边缘检测效果。关键词：图像处理；边缘检测；Canny算子；滤波；自适应阈值东华理工大学长江学院毕业设计（论文）摘要ABSTRACTDigitalimageedgedetectionplaysanimportpartinimageanalysis,suchasimagesegmentation,interestedregionrecognitionandregionshapeextraction.Andit’sanimportmethodinimagefeatureextractionofimagerecognition.Theedgeincludesthevaluableinfotmationoftheimagewhichcanbeuseinimageunderstandingandanalysis.Andthroughedgedetection,wecangreatlyreducethecalculationofimageanalysisandprocessinginthefollowingstep.Usually,thefirststepofimageunderstandingandanalysisisedgedetection,andithasbeenthemostactivetopicinthemachinevisionresearchfield,alsoitplaysanimportpartinengineeringapplication.Mostofthetraditionaledgedetectionalgorithms,suchasRoberts,Sobel,Prewitt,Kirsch,Laplacian,justconstructanedgedetectionalgorithmwithasmallneighborhoodineachpixeloftheoriginalimage,andthencarryoutwithfirstdifferentialorseconddifferentialoperatorinordertoobtainthemaximumgradientorthezero-crossingpointofthesecondderivative,finallyselectanappropriatethresholdtoextracttheedge.Butthesealgorithmssharethesameshortcomings,forexample,theyaresensitivetonoise,theycan’tselectthresholdadaptively,andthedetectionresultsarenotsowell.Inthispaper,wedoadeepresearchontheedgedetectiontheoryandalgorithm,baseonanalyzingthetraditionaledgedetectionalgorithmindetail,wefocusonCannyalgorithm,combinedwithMTMalgorithmandOtsualgorithmtoimprovethefilteringmethodanddualthresholdselectionmethodinCannyalgorithm,andwecalltheimprivedalgorithmCMOforshort.Finally,weuseMATLAB7.0toimplementCMOalgorithm,andtheexperimentresultsshowthatCMOalgorithmcangetbetterresultsthantraditionalCannyalgorithm.Keywords:ImageProcessing;EdgeDetction;Canny;Filtering;AdaptiveThreshold东华理工大学长江学院毕业设计（论文）摘要目录绪论................................................................................................................................11.数字图像处理..........................................................................................................11.1数字图像处理的发展.........................................................................................11.2数字图像处理的应用.........................................................................................11.3数字图像边缘定义及边缘提取方法概述.........................................................41.4目前边缘检测存在的问题.................................................................................61.5本文主要研究工作.............................................................................................62.图像滤波..................................................................................................................72.1图像噪声的定义.................................................................................................72.2图像噪声的来源.................................................................................................72.3图像噪声的滤除.................................................................................................82.3.1领域平均法............................................................................................................................82.3.2加权平均法..........................................................................................................................102.3.3中值滤波..............................................................................................................................112.3.4空域低通滤波......................................................................................................................123.传统边缘检测算法的研究与分析........................................................................143.1图像边缘检测方法分类...................................................................................143.1.1基于空间域上微分算子的边缘提取方法..........................................................................143.1.2基于图像滤波的边缘提取方法..........................................................................................153.2图像边缘评价指标...........................................................................................163.3尺度对性能指标的影响...................................................................................173.4经典边缘检测方法综述...................................................................................183.4.1基于灰度直方图的边缘检测..............................................................................................183.4.2ROBERTS算子..................................................................................................193.4.3Sobel算子............................................................................