ImageProcessing3-ImageEnhancement(HistogramProcess

CourseWebsite:(HistogramProcessing)2of32ComeToTheLABS!Day:WednesdayTime:9:00–11:00Room:AungierSt.1-005WewillstartbygettingtogripswiththebasicsofScilab–LabdetailsavailableatWebCTShortly,therewillbeaScilabassignmentwhichwillcounttowardsyourfinalmark3of32ContentsOverthenextfewlectureswewilllookatimageenhancementtechniquesworkinginthespatialdomain:–Whatisimageenhancement?–Differentkindsofimageenhancement–Histogramprocessing–Pointprocessing–Neighbourhoodoperations4of32ANoteAboutGreyLevelsSofarwhenwehavespokenaboutimagegreylevelvalueswehavesaidtheyareintherange[0,255]–Where0isblackand255iswhiteThereisnoreasonwhywehavetousethisrange–Therange[0,255]stemsfromdisplaytechnologesFormanyoftheimageprocessingoperationsinthislecturegreylevelsareassumedtobegivenintherange[0.0,1.0]5of32WhatIsImageEnhancement?ImageenhancementistheprocessofmakingimagesmoreusefulThereasonsfordoingthisinclude:–Highlightinginterestingdetailinimages–Removingnoisefromimages–Makingimagesmorevisuallyappealing6of32ImageEnhancementExamplesImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)7of32ImageEnhancementExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)8of32ImageEnhancementExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)9of32ImageEnhancementExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)10of32Spatial&FrequencyDomainsTherearetwobroadcategoriesofimageenhancementtechniques–Spatialdomaintechniques•Directmanipulationofimagepixels–Frequencydomaintechniques•ManipulationofFouriertransformorwavelettransformofanimageForthemomentwewillconcentrateontechniquesthatoperateinthespatialdomain11of32ImageHistogramsThehistogramofanimageshowsusthedistributionofgreylevelsintheimageMassivelyusefulinimageprocessing,especiallyinsegmentationGreyLevelsFrequencies12of32HistogramExamplesImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)13of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)14of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)15of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)16of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)17of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)18of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)19of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)20of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)21of32HistogramExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)22of32HistogramExamples(cont…)AselectionofimagesandtheirhistogramsNoticetherelationshipsbetweentheimagesandtheirhistogramsNotethatthehighcontrastimagehasthemostevenlyspacedhistogramImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)23of32ContrastStretchingWecanfiximagesthathavepoorcontrastbyapplyingaprettysimplecontrastspecificationTheinterestingpartishowdowedecideonthistransformationfunction?24of32HistogramEqualisationSpreadingoutthefrequenciesinanimage(orequalisingtheimage)isasimplewaytoimprovedarkorwashedout(褪色)imagesTheformulaforhistogramequalisationisgivenwhere–rk:inputintensity–sk:processedintensity–k:theintensityrange(e.g0.0–1.0)–nj:thefrequencyofintensityj–n:thesumofallfrequencies)(kkrTskjjrrp1)(kjjnn125of32EqualisationTransformationFunctionImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)26of32EqualisationExamplesImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)127of32EqualisationTransformationFunctionsThefunctionsusedtoequalisetheimagesinthepreviousexampleImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)28of32EqualisationExamplesImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)229of32EqualisationTransformationFunctionsThefunctionsusedtoequalisetheimagesinthepreviousexampleImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)30of32EqualisationExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)3431of32EqualisationExamples(cont…)ImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)3432of32EqualisationTransformationFunctionsThefunctionsusedtoequalisetheimagesinthepreviousexamplesImagestakenfromGonzalez&Woods,DigitalImageProcessing(2002)33of32SummaryWehavelookedat:–Differentkindsofimageenhancement–Histograms–HistogramequalisationNexttimewewillstarttolookatpointprocessingandsomeneighbourhoodoperations