任務8-Canny技術找出圖像輪廓2024-07-20Task8-CannytechnologytofindtheimageoutlineCanny算法的應用ApplicationofCannyalgorithmCanny算法的原理TheprincipleofCannyalgorithm圖像輪廓檢測的實現(xiàn)步驟Theimplementationstepsofimagecontourdetection代碼實現(xiàn)Codeimplementation總結Summary目錄CATALOGUEPART01Canny算法的ApplicationofCannyalgorithmCanny算法在圖像處理中的作用TheroleofCannyalgorithminimageprocessingCanny算法是計算機視覺領域中常用的邊緣檢測方法，它能夠準確地找出圖像中的輪廓區(qū)域。Cannyalgorithmisacommonlyusededgedetectionmethodinthefieldofcomputervision,whichcanaccuratelyfindoutthecontourregionintheimage.Canny算法的優(yōu)勢AdvantagesoftheCannyalgorithmCanny算法具有抗噪聲能力強、定位準確等特點，適用于各種復雜的圖像環(huán)境。Cannyalgorithmhasthecharacteristicsofstronganti-noiseability,accuratepositioning,etc.,whichissuitableforavarietyofcompleximageenvironments.Canny算法的應用TheapplicationofCannyalgorithmPART02Canny算法的原理HowtheCannyalgorithmworksCanny算法是一種基于梯度的邊緣檢測算法，具有抗噪聲能力強、定位準確等特點。Cannyalgorithmisagradient-basededgedetectionalgorithm,whichhasthecharacteristicsofstronganti-noiseabilityandaccuratepositioning.Canny算法CannyalgorithmCanny算法的主要步驟包括高斯濾波、計算梯度幅值和方向、非極大值抑制以及雙閾值處理。ThemainstepsoftheCannyalgorithmincludeGaussianfiltering,calculatingthegradientamplitudeanddirection,non-maximumsuppression,anddoublethresholdprocessing.邊緣檢測EdgeDetection高斯濾波使用一個高斯核對圖像進行卷積操作，能夠有效地減少噪聲的影響。GaussianfilteringusesaGaussiancheckimageforconvolutionoperation,whichcaneffectivelyreducetheinfluenceofnoise.高斯濾波GaussianfilteringCanny算法概述Cannyalgorithmoverview010203Canny算法概述Cannyalgorithmoverview梯度計算GradientComputing對平滑后的圖像計算梯度，常用的梯度算子有Sobel算子和Prewitt算子。SobeloperatorandPrewittoperatorarecommonlyusedtocalculatethegradientofthesmoothedimage.非極大值抑制Non-maximumsuppression根據(jù)梯度方向，對圖像進行非極大值抑制，選擇局部最大值作為邊緣像素。Accordingtothegradientdirection,non-maximumsuppressionisappliedtotheimage,selectingthelocalmaximumastheedgepixel.雙閾值處理Doublethresholdprocessing將梯度幅值進行閾值分割，得到強邊緣和弱邊緣，便于后續(xù)的邊緣連接。Thegradientamplitudeissegmentedbythresholdtogetstrongedgeandweakedge,whichisconvenientforsubsequentedgeconnection.邊緣連接Edgejoining根據(jù)強邊緣像素的鄰域信息，將弱邊緣連接到強邊緣，形成閉合的邊緣輪廓。Accordingtotheneighborhoodinformationofthestrongedgepixels,theweakedgeisconnectedtothestrongedgetoformaclosededgeoutline.高斯核GaussianKernel高斯核是一個重要的概念，它決定了高斯濾波的效果。核的大小和標準差需要根據(jù)具體情況進行調(diào)整。GaussiankernelisanimportantconceptthatdeterminestheeffectofGaussianfiltering.Thesizeandstandarddeviationofthekernelneedtobeadjustedaccordingtothespecificsituation.高斯濾波Gaussianfiltering高斯濾波使用一個高斯核對圖像進行卷積操作，能夠有效地減少噪聲的影響。GaussianfilteringusesaGaussiancheckimageforconvolutionoperation,whichcaneffectivelyreducetheinfluenceofnoise.平滑圖像Smoothingtheimage高斯濾波旨在平滑圖像并抑制噪聲，從而為后續(xù)的邊緣檢測步驟提供更好的基礎。Gaussianfilteringisdesignedtosmooththeimageandsuppressnoise,thusprovidingabetterbasisforsubsequentedgedetectionsteps.高斯濾波Gaussianfiltering對平滑后的圖像計算梯度，常用的梯度算子有Sobel算子和Prewitt算子。SobeloperatorandPrewittoperatorarecommonlyusedtocalculatethegradientofthesmoothedimage.梯度計算GradientcalculationSobel算子和Prewitt算子是兩種常用的梯度算子，通過對圖像在水平和垂直方向上應用這些算子進行卷積操作，得到圖像的梯度幅值和方向。SobeloperatorandPrewittoperatoraretwocommonlyusedgradientoperators.Thegradientamplitudeanddirectionoftheimagecanbeobtainedbyapplyingtheseoperatorstotheconvolutionoperationonthehorizontalandverticaldirectionsoftheimage.梯度算子Gradientoperators梯度計算Gradientcalculation非極大值抑制Non-maximumsuppression插值操作Interpolationoperation在梯度方向上進行插值操作，選擇局部最大值作為邊緣像素，并抑制非邊緣像素，能夠更好地定位圖像中的邊緣區(qū)域。Interpolatinginthegradientdirection,selectingthelocalmaximumastheedgepixel,andsuppressingthenon-edgepixel,enablesbetterpositioningoftheedgeregionintheimage.非極大值抑制Non-maximumsuppression根據(jù)梯度方向，對圖像進行非極大值抑制，通過在梯度方向上進行插值操作，選擇局部最大值作為邊緣像素，并抑制非邊緣像素。Accordingtothegradientdirection,non-maximumsuppressionisappliedtotheimagebyperforminganinterpolationoperationinthegradientdirection,selectingthelocalmaximumastheedgepixel,andsuppressingthenon-edgepixel.雙閾值處理Doublethresholdprocessing將梯度幅值進行閾值分割，得到強邊緣和弱邊緣，通常將梯度幅值大于高閾值的像素判定為強邊緣，將梯度幅值介于低閾值和高閾值之間的像素判定為弱邊緣。Thegradientamplitudeisdividedbythresholdtogetstrongedgeandweakedge,usuallythepixelwhosegradientamplitudeisgreaterthanthehighthresholdisjudgedasastrongedge,andthepixelwhosegradientamplitudeisbetweenthelowthresholdandthehighthresholdisjudgedasaweakedge.梯度幅值分割Gradientamplitudesegmentation通過閾值分割，能夠將圖像中的梯度幅值進行區(qū)分，得到強邊緣和弱邊緣，從而更好地提取出圖像中的輪廓信息。Throughthresholdsegmentation,thegradientamplitudeintheimagecanbedistinguishedtogetstrongedgeandweakedge,soastobetterextractthecontourinformationintheimage.雙閾值處理Doublethresholdprocessing邊緣連接Edgejoin根據(jù)強邊緣像素的鄰域信息，將弱邊緣連接到強邊緣，形成閉合的邊緣輪廓。Accordingtotheneighborhoodinformationofthestrongedgepixels,theweakedgeisconnectedtothestrongedgetoformaclosededgeoutline.鄰域信息Neighborhoodinformation利用鄰域信息，將弱邊緣連接到強邊緣，能夠更好地形成閉合的邊緣輪廓。Byusingtheneighborhoodinformation,theweakedgeisconnectedtothestrongedge,andtheclosededgeprofilecanbebetterformed.邊緣連接EdgeJoiningPART03圖像輪廓檢測的實現(xiàn)步驟Theimplementationstepsofimagecontourdetection在Python中，可以使用OpenCV庫等進行圖像處理，通過導入相關的庫，可以使用Canny函數(shù)進行圖像邊緣檢測。InPython,youcanuseOpenCVlibraryforimageprocessing,byimportingrelatedlibraries,youcanuseCannyfunctionforimageedgedetection.導入OpenCV庫ImportOpenCVlibraryOpenCV庫中的Canny函數(shù)是一種常用的邊緣檢測算法，通過導入OpenCV庫，可以方便地使用Canny函數(shù)進行圖像邊緣檢測。CannyfunctioninOpenCVlibraryisacommonlyusededgedetectionalgorithm,byimportingOpenCVlibrary,youcaneasilyuseCannyfunctionforimageedgedetection.Canny函數(shù)進行邊緣檢測Cannyfunctionforedgedetection導入相關的庫Importtherelevantlibraries讀取圖像Readimages使用圖像讀取函數(shù)將圖像讀入內(nèi)存，方便進行后續(xù)的圖像處理操作。Usetheimagereadfunctiontoreadtheimageintomemoryforeasysubsequentimageprocessingoperations.轉換灰度圖像Convertgrayscaleimages讀取圖像并轉化為灰度圖像Readtheimageandtransformitintoagrayscaleimage將彩色圖像轉換為灰度圖像，是因為Canny算法是基于灰度圖像的操作，轉換后能更好地適應算法要求。ThecolorimageisconvertedtoagrayscaleimagebecausetheCannyalgorithmisbasedontheoperationofthegrayscaleimage,whichcanbetteradapttothealgorithmrequirementsafterconversion.0102高斯濾波Readimages對轉化后的灰度圖像進行高斯濾波操作，以平滑圖像并抑制噪聲。Usetheimagereadfunctiontoreadtheimageintomemoryforeasysubsequentimageprocessingoperations.高斯濾波器Convertgrayscaleimages高斯濾波使用一個高斯核對圖像進行卷積操作，它能夠有效地減少噪聲的影響。ThecolorimageisconvertedtoagrayscaleimagebecausetheCannyalgorithmisbasedontheoperationofthegrayscaleimage,whichcanbetteradapttothealgorithmrequirementsafterconversion.高斯濾波ReadtheimageandtransformitintoagrayscaleimageVS對平滑后的圖像計算梯度，常用的梯度算子有Sobel和Prewitt。Forcalculatingthegradientofthesmoothedimage,commongradientoperatorsareSobelandPrewitt.梯度幅值和方向Gradientamplitudeanddirection通過卷積操作得到圖像的梯度幅值和方向，用于后續(xù)邊緣檢測。Thegradientamplitudeanddirectionoftheimageareobtainedthroughtheconvolutionoperationforsubsequentedgedetection.梯度計算Gradientcalculation計算梯度幅值和方向Readtheimageandtransformitintoagrayscaleimage根據(jù)梯度方向，對圖像進行非極大值抑制，選擇局部最大值作為邊緣像素。Accordingtothegradientdirection,non-maximumsuppressionisappliedtotheimageandthelocalmaximumvalueisselectedastheedgepixel.非極大值抑制Non-maximumsuppression在梯度方向上進行插值操作，抑制非邊緣像素，提高邊緣檢測的準確性。Theinterpolationoperationiscarriedoutinthegradientdirectiontosuppressnon-edgepixelsandimprovetheaccuracyofedgedetection.插值操作Interpolatingoperation非極大值抑制Non-maximumsuppression雙閾值處理Doublethresholdprocessing邊緣判定EdgeDecision像素梯度幅值大于高閾值的為強邊緣，介于低閾值和高閾值之間的為弱邊緣。Thepixelgradientamplitudelargerthanthehighthresholdisastrongedge,andthepixelgradientamplitudebetweenthelowthresholdandthehighthresholdisaweakedge.雙閾值處理Doublethresholdprocessing將梯度幅值進行閾值分割，得到強邊緣和弱邊緣。Thegradientamplitudeissegmentedbythresholdtogetstrongedgeandweakedge.強邊緣像素Strongedgepixels根據(jù)強邊緣像素的鄰域信息，將弱邊緣連接到強邊緣，形成閉合的邊緣輪廓。Basedontheneighborhoodinformationofthestrongedgepixels,connecttheweakedgetothestrongedgetoformaclosededgeoutline.弱邊緣連接WeakedgeJoining弱邊緣連接強邊緣，需找到強邊緣像素的鄰域，將符合條件的弱邊緣像素連接到強邊緣上。Toconnecttheweakedgetothestrongedge,itisnecessarytofindtheneighborhoodofthestrongedgepixelandconnectthequalifiedweakedgepixeltothestrongedge.邊緣輪廓提取Edgecontourextraction通過連接弱邊緣和強邊緣，可以提取出完整的邊緣輪廓，用于后續(xù)的目標識別和圖像分割。Byconnectingweakedgesandstrongedges,acompleteedgecontourcanbeextractedforsubsequenttargetrecognitionandimagesegmentation.邊緣連接EdgeJoining使用圖像顯示函數(shù)將原始圖像進行顯示，以便與檢測到的邊緣圖像進行對比。Usetheimagedisplayfunctiontodisplaytheoriginalimageforcomparisonwiththedetectededgeimage.原始圖像顯示RawimageDisplay使用圖像顯示函數(shù)將檢測到的邊緣圖像進行顯示，突出輪廓檢測的結果。Usetheimagedisplayfunctiontodisplaythedetectededgeimageandhighlighttheresultofthecontourdetection.邊緣圖像顯示Edgeimagedisplay通過原始圖像和邊緣圖像的對比顯示，可以清晰地看到輪廓檢測的效果。Theeffectofcontourdetectioncanbeclearlyseenthroughthecontrastdisplayoftheoriginalimageandtheedgeimage.對比顯示Contrastdisplay顯示結果ShowresultsPART04代碼實現(xiàn)CodeImplementation代碼實現(xiàn)Codeimplementation導入所需的庫Importtherequiredlibrary在Python中，使用OpenCV庫進行圖像處理，通過導入相關庫，使用Canny函數(shù)進行圖像邊緣檢測。InPython,usetheOpenCVlibraryforimageprocessing,andusetheCannyfunctionforimageedgedetectionbyimportingrelatedlibraries.讀取并轉換圖像Readandconverttheimage對圖像進行高斯濾波Gaussianfiltertheimage使用圖像讀取函數(shù)將圖像讀入內(nèi)存，將彩色圖像轉換為灰度圖像，因為Canny算法是基于灰度圖像的操作。TheimageisreadintomemoryusingtheimagereadfunctiontoconvertthecolorimagetoagrayscaleimagebecausetheCannyalgorithmisanoperationbasedongrayscaleimages.對轉化后的灰度圖像進行高斯濾波操作，以平滑圖像并抑制噪聲。TheconvertedgrayscaleimageisoperatedbyGaussianfilteringtosmooththeimageandsuppressthenoise.計算梯度幅值和方向Calculatethegradientamplitudeanddirection使用Canny函數(shù)對平滑后的圖像進行梯度計算，得到圖像的梯度幅值和方向。TheCannyfunctionisusedtocalculatethegradientofthesmoothedimagetoobtainthegradientamplitudeanddirectionoftheimage.代碼實現(xiàn)Codeimplementation進行非極大值抑制Donon-maximumsuppression根據(jù)梯度方向，對圖像進行非極大值抑制操作，選擇局部最大值作為邊緣像素。Accordingtothegradientdirection,performanon-maximumsuppressoperationontheimage,selectingthelocalmaximumastheedgepixel.對梯度幅值進行雙閾值處理Thegradientamplitudeisprocessedbydoublethreshold對梯度幅值進行雙閾值處理，將梯度幅值大于高閾值的像素判定為強邊緣，介于低閾值和高閾值之間的像素判定為弱邊緣。Thegradientamplitudeisdouble-thresholdprocessed,andthepixelswhosegradientamplitudeislargerthanthehighthresholdarejudgedasstrongedges,andthosebetweenthelowthresholdandthehighthresholdarejudgedasweakedges.根據(jù)強邊緣像素的鄰域信息，將弱邊緣連接到強邊緣，形成閉合的邊緣輪廓。Accordingtotheneighborhoodinformationofthepixelswithstrongedges,theweakedgesareconnectedtothestrongedgestoformaclosededgeprofile.根據(jù)鄰域信息連接邊緣Connectedgesbasedonneighborhoodinformation使用圖像顯示函數(shù)將圖像進行顯示，可以將原始圖像和檢測到的邊緣圖像進行對比顯示。Usingtheimagedisplayfuncti