Chapter 1 - Introduction

《Chapter 1 - Introduction》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、§1-IntroductionDr.DirkSchniedersDepartmentofComputerScienceTheUniversityofHongKongIntroduction★WhatisImageProcessing(IP)?★Improvementofpictorialinformationforhumaninterpretation★e.g.imageenhancement,imagerestoration★Processingofimagedataforstorage,transmission,andr

2、epresentationforautonomousmachineperception★e.g.imagecompression,imagesegmentation★DigitalIPreferstoprocessingofdigitalimagesbymeansofacomputer2013-2014COMP75022IPExample12013-2014COMP75023IPExample22013-2014COMP75024IPExample32013-2014COMP75025Topics(thiscourse)★D

3、igitalImageFundamentals★SpatialImageEnhancement★FrequencyDomainEnhancement★ImageRestoration2013-2014COMP75026IntroductionWhatisComputerVision(CV)?AbranchofartificialintelligenceAnalyzeandunderstandimagesEmulationofhumanvisionsystem,includinglearning,beingable

4、tomakeinference,andtakingactionsbasedonvisualinputse.g.navigation,manipulation,recognition2013-2014COMP75027CVExample12013-2014COMP75028CVExample22013-2014COMP75029CVExample32013-2014COMP750210Topics(thiscourse)★FeatureExtraction★CameraModel★CameraCalibration★Ster

5、eoVision2013-2014COMP750211IPvsCVThereisacontinuumfromimageprocessingatoneendtocomputervisionattheotherNoclear-cutboundariesinthiscontinuumTheareaImageAnalysis(orImageUnderstanding)isinbetweenimageprocessingtocomputervisionImageImageComputerProcessingAnalysisVis

6、ion2013-2014COMP750212LevelsofProcessesTypically,threetypesofprocessescanbeclassifiedinthiscontinuum:low-levelprocesses,mid-levelprocesses,andhigh-levelprocessesImageComputerProcessingVisionlow-levelmid-levelhigh-levelprocessesprocessesprocesses2013-2014COMP750

7、213Low-LevelProcessesLow-levelprocessesPrimitiveoperationse.g.noiseremoval,contrastenhancement,imagesharpeningTypically,bothinputsandoutputsareimageslow-levelimageimageprocess2013-2014COMP750214Mid-LevelProcessesMid-levelprocessesHigherleveltaskse.g.imagesegme

8、ntation,imagedescription,classificationandrecognitionInputsaregenerallyimagesOutputsareattributesextractedfromtheimagese.g.edges,contours,ident