计算机视觉课程1.ppt

《计算机视觉课程1.ppt》由会员分享，可在线阅读，更多相关《计算机视觉课程1.ppt（17页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、Teacher:Prof.LiuYuncai 计算机视觉课程计算机视觉课程COMPUTERVISIONReferencebooks:1.MachineVision,byRamerhJian,etal(MajorReference)2.ComputerVision,byDamaH.Ballard,etal.3.Vision,ByDavidMarr.4.IntroductoryTechniquesfor3-DComputerVision,byEmanueleTrucco.5.ComputerVisionandImageProcessing,byScattE.Umbaugh.6.Algorithmf

2、orImageProcessingandComputerVision,byJ.R.Parker.7.机器视觉，贾云得 编著8.计算机视觉的理论和实践，李介谷著9.计算机视觉计算机理论与算法基础，马颂德Ruleofgrading:Homework:10%Examination:60%Project:30%ImportantDeadlines:Handinprojectproposal:May11,2010Handinprojectreport:July6,2010AllwritteninChinese(中文书写)FinalExamination:July6,2010Course 1 Introd

3、uctionCourse 1Introduction1.WhatisComputerVisionComputer vision aims to recover usefulinformationabouta(3D)scenefromits2Dprojections(images).n3Ddepth/3Dstructuren3Dmotionanalysisn3Dsurfacesandorientationnstatusof3Dobjects,soonnmeaningofactionsof3Dobjects(cognition)Everytaskistoughsinceitistobeunders

4、toodfromflatimages!Computervisionisalsocalledas:MachinevisionRobotvisionImageunderstandingSceneanalysis(inearlyyears)课程内容：课程内容：1)绪论(Introduction)2)图象滤波(ImageFiltering)3)二进制图象处理(BinaryImageProcessing)4)区域(Region)5)边缘检测(EdgeDetection)6)立体视觉(Stereo)7)运动的理解与估值(Motion)8)轮廓(Contours)9)纹理(Texture)10)图象光度学(

5、Shading)11)光流场(OpticFlows)12)系统校准(Calibration)13)曲线与曲面(CurvesandSurfaces)14)动态视觉(DynamicVision)15)三维识别(ObjectRecognition)2.RelatedfieldsimageprocessingcomputervisionpatternrecognitionartificialIntelligenceneuralnetworkpsychophysics3.PerspectiveProjectionpin-holecameramodeln projectioncenter(i.e.,cam

6、era)isattheoriginofscenecoordinatesystemwithopticaxisofcamerapointingtoz-axisdirectionofcoordinatesnoriginofimagecoordinatesystemislocatedat(0,0,f)ofscenecoordinatesystem(fisfocuslengthofthecamera),theX-andY-axesofimagecoordinatesareparallelwithx-andy-axesofscenecoordinates,respectivelyncameradoesno

7、thaveanylensdistortions.xoyoXYP(X,Y)P(x,y,z)LlnP0focus length=frRz(1)Imagepoint(2)Image Line(3)3D line l is projected onto image plane through projecting plane to get image line L:Let we use a vector to represent the image line L of l.Since We have,orIt is the equation of projecting plane,the normal

8、 is .Thus,we will use the normal of the projecting plane of 3D line to denote image plane by4.DigitalImagenSampling:digitalizeinimageplane(directions)say,andsoon.nquantization:digitalizeinimageintensity,say8bits(256levels).foranimageshavingonly2graylevels,wecallitasbinaryimage.OjiOXYImage of 64 gray leveslBinary image