四川理工学院毕业设计（论文）边缘检测技术在汽车牌照自动识别监控系统中的应用四川理工学院自动化与电子信息学院二00九年六月摘要车牌识别是智能交通中关键技术之一。车牌识别系统一般包括车辆图像采集、车牌定位、字符分割和字符识别四个模块。本文在简单介绍车牌识别的四个模块的基础上，主要对车牌边缘检测进行了讨论，分析了Roberts算子、Sobel算子、canny算子、拉普拉斯(Laplacian)算子和Marr边缘检测算子。Roberts算子检测定位精度比较高，但对噪声敏感。Sobel算子可以产生较好的边缘效果，但是对噪声具有平滑作用，减小了对噪声的敏感性。但是，Sobel边缘检测算子也检测出了一些伪边缘，使得边缘比较粗，降低了检测定位精度。canny算子能在噪声抑制和边缘检测之间取得较好的平衡，并能产生较细的边缘效果。故本文采用canny算子对车牌的边缘进行了提取，仿真结果表明了该算子能较好的检测出图片的边缘信息。关键词：车牌识别，车牌定位，字符分割，边缘检测ABSTRACTThecarlicenserecognitionisoneofkeytechnologiesintheintelligenttransportation.Thecarlicenserecognitionsystemincludesvehiclesimagegathering,thecarlicenselocalizationgenerally,thecharacterdivisionandthecharacterrecognitionfourmodules.Thisarticleinthesimpleintroductioncarlicenserecognitionfourmodulefoundations,mainlyhascarriedonthediscussiontothecarlicenseedgeexamination,hasanalyzedtheRobertsoperator,theSobeloperator,thecannyoperator,theLaplacian(Laplacian)operatorandtheMarredgeexaminationoperator.TheRobertsoperatorexaminationpointingaccuracyquiteishigh,butissensitivetothenoise.TheSobeloperatormayproducethegoodedgeresults,buthasthesmoothingeffecttothenoise,reducedtothenoisesensitivity.But,theSobeledgeexaminationoperatoralsoexaminedsomefalseedges,causedtheedgequitetobethick,reducedtheexaminationpointingaccuracy.Thecannyoperatorcanobtainsthegoodbalancebetweenthenoiseabatementandtheedgeexamination,Itcanproducethethinedgeresults.Thereforethisarticleusedthecannyoperatortocarryontheextractiontothecarlicenseedge,thesimulationresulthasindicatedthisoperatortobeablethegoodexaminationpictureedgeinformation.Keywords:carlicenserecognition,carlicenselocalization,characterdivision,edgeexamination3目录摘要............................................................IABSTRACT.........................................................II第1章引言......................................................11.1课题背景及研究意义........................................11.2国内外研究现状............................................21.2.1汽车牌照识别国内外研究现状..........................21.2.2边缘检测算法国内外研究现状..........................31.3我国汽车牌照识别的难点....................................51.4课题的主要研究内容........................................6第2章汽车图像的采集.............................................72.1常用的车辆检测方法.........................................72.2运动车辆检测算法概述.......................................72.2.1光流场法.............................................72.2.2帧差法...............................................82.2.3背景消减法...........................................82.3背景图像的提取与更新.......................................92.3.1背景图像的提取.......................................92.3.2背景图像的更新.......................................92.4有无车辆判定..............................................10第3章车牌定位..................................................113.1现行汽车牌照的规格.......................................113.1.1现行汽车牌照的...
