电子信息专业英语-11

信息科学与电子工程专业英语TechnicalEnglishForInformationScienceandElectronicEngineering信息科学与电子工程专业英语Unit11DigitalImageProcessing信息科学与电子工程专业英语PartITwo-DimensionalDigitalImages3NewWordsparameter参数Intensity强度multi-spectral多光谱的manipulate处理，操作introductory介绍性的，引导性的function函数amplitude幅度coordinate坐标regions-of-interest感兴趣区motionblur运动模糊rendition表现，渲染render表现，渲染realnumber实数integer整数quantization量化discrete离散的photon光子inherently内在地，本质上magneticresonance磁共振phase相位，相角4NewWordsMRI磁共振成像sampling采样digitization数字化intersection交叉点pixel像素impinge撞击，射到sensor传感器，感光器monochromatic单色的round四舍五入convolution卷积outset开头，最初perceptual感官的51Imagesaretwo-dimensionalsignalsrepresentingspatialdistributionofsomephysicalparameters,typicallylightintensity,andmoregenerallyanyformofenergy.Motionpicturesandmulti-spectralremotesensingimages,forexample,arethreeorhigherdimensionalsignals.多光谱的遥感图像61Moderndigitaltechnologyhasmadeitpossibletomanipulatemulti-dimensionalsignalswithsystemsthatrangefromsimpledigitalcircuitstoadvancedparallelcomputers.1现代数字技术使得处理多维信号成为可能，所使用的系统可从简单的数字电路到先进的并行计算机。71Thegoalofthismanipulationcanbedividedintothreecategories:Imageprocessing:imageinimageoutImageanalysis:imageinmeasurementsoutImageunderstanding:imageinhigh-leveldescriptionout82Wewillfocusonthefundamentalconceptsofimageprocessing.Spacedoesnotpermitustomakemorethanafewintroductoryremarksaboutimageanalysisandimageunderstanding.Further,wewillrestrictourselvestotwo-dimensional(2D)imageprocessingalthoughmostoftheconceptsandtechniquesthataretobedescribedcanbeextendedeasilytothreeormoredimensions.2另外我们将限于讨论二维图像处理，尽管将要描述的大部分概念和方法很容易扩展到三维或更高的维数。93Webeginwithcertainbasicdefinitions.Animagedefinedinthe“realworld”isconsideredtobeafunctionoftworealvariables,forexample,a(x,y)withaastheamplitude(e.g.brightness)oftheimageattherealcoordinateposition(x,y).Animagemaybeconsideredtocontainsub-imagessometimesreferredtoasregions-of-interest,ROIs,orsimplyregions.Thisconceptreflectsthefactthatimagesfrequentlycontaincollectionsofobjectseachofwhichcanbethebasisforaregion.103Inasophisticatedimageprocessingsystemitshouldbepossibletoapplyspecificimageprocessingoperationstoselectedregions.Thusonepartofanimagemightbeprocessedtosuppressmotionblurwhileanotherpartmightbeprocessedtoimprovecolorrendition.3于是图像的一部分可以进行抑制运动模糊的处理，而另一部分可以作改进色彩渲染的处理。114Theamplitudesofagivenimagewillalmostalwaysbeeitherrealnumbersorintegernumbers.Thelatterisusuallyaresultofaquantizationprocessthatconvertsacontinuousrange,say,between0and100%,toadiscretenumberoflevels.4后者通常是量化的结果，将一个连续的范围如0到100%转换为离散的（灰度）级。124Incertainimage-formingprocesses,however,thesignalmayinvolvephotoncountingwhichimpliesthattheamplitudewouldbeinherentlyquantized.Inotherimageformingprocedures,suchasmagneticresonanceimaging,thedirectphysicalmeasurementyieldsacomplexnumberintheformofarealmagnitudeandarealphase.Wewillconsideramplitudesasrealsorintegersunlessotherwiseindicated.这意味着幅度本来是量化的135Adigitalimagea[m,n]describedina2Ddiscretespaceisderivedfromananalogimagea(x,y)ina2Dcontinuousspacethroughasamplingprocessthatisfrequentlyreferredtoasdigitization.5一幅二维离散空间中的数字图像a[m,n]是由二维连续空间中的模拟图像a(x,y)通过采样得到的。采样过程常称为数字化。145Themathematicsofthatsamplingprocesswillbedescribedlater.Fornowwewilllookatsomebasicdefinitionsassociatedwiththedigitalimage.TheeffectofdigitizationisshowninFigure11.1.156The2Dcontinuousimagea(x,y)isdividedintoNrowsandMcolumns.Theintersectionofarowandacolumnistermedapixel.Thevalueassignedtotheintegercoordinates[m,n]withm=0,1,2,...,M1andn=0,1,2,...,N1isa[m,n].166Infact,inmostcasesa(x,y),whichwemightconsidertobethephysicalsignalthatimpingesonthefaceofa2Dsensor,isactuallyafunctionofmanyvariablesincludingdepth,color,andtime.6Unlessotherwisestated,wewillconsiderthecaseof2D,monochromatic,staticimagesinthistext.大多数情况下，我们认为是射到二维传感器表面的物理信号a(x,y)实际上是许多变量的函数，包括深度、颜色、时间。177TheimageshowninFigure11.1hasbeendividedintoN=16rowsandM=16columns.Thevalueassignedtoeverypixelistheaveragebrightnessinthepixelroundedtothenearestintegervalue.赋予每个像素的值是像素的平均亮度，其四舍五入到最近的整数187Theprocessofrepresentingtheamplitudeofthe2DsignalatagivencoordinateasanintegervaluewithLdifferentgraylevelsisusuallyreferredtoasamplitudequantizationorsimplyquantization.7将二维信号在给定坐标位置的幅度表示成具有L个不同灰度级的整数值的过程通常称为幅度量化，或简称量化。198Certaintoolsarecentraltotheprocessingofdigitalimages.Theseincludemathematicaltoolssuchasconvolution,Fourieranalysis,andstatisticaldescriptions,andmanipulativetoolssuchaschaincodesandruncodes.Herewepresentthesetoolswithoutanyspecificmotivation.Themotivationwillfollowinlatersections.数学工具，如：卷积、傅立叶分析以及统计描述，和处理工具，如：链式码与游程码209Manyimageprocessingapplicationsareintendedtoproduceimagesthataretobeviewedbyhumanobserversasopposedto,say,automatedindustrialinspection.Itisthereforeimportanttounderstandthecharacteristicsandlimitationsofthehumanvisualsystem―tounderstandthe“receiver”ofthe2Dsignals.许多图像处理应用是要产生被人观察的图像，而不是比方说为了自动的工业检验219Attheoutsetitisimportanttorealizethat1)thehumanvisualsystemisnotwellunderstood,2)noobjectivemeasureexistsforjudgingthequalityofanimagethatcorrespondstohumanassessmentofimagequality,and3)the“t