基于多分辨率的圖像拼接技術(shù).pdf

1、分類號UDC丁foQT7密級盤自Y86152學(xué)校代碼：10561學(xué)號：200235000201028華南理工大學(xué)碩士學(xué)位論文基于多分辨率的圖像拼接技術(shù)鄭欽指導(dǎo)教師：鏊叢鋈副麴握魚王曼籃：墾堂暄申請學(xué)位級別：亟專業(yè)名稱：逼籃皇信息丕統(tǒng)論文提交日期：!QQ：論文答辯日期：21Qi：』靴黔靴糍雨掣堡旌趣答辯委員會主席：詐終升段書L論文評閱人：華南理丁大學(xué)碩十學(xué)位論文ABSTRACTImagemosaicisthecrucialstepofim

2、ageprocessingItisanactiveareaofresearchinthefieldsofphotogrammetry，theconstructionofvirtualenvironments，satellitephotographs，computervision，imageprocessing，computergraphicsandmedicalimageeta1Actually，Imagemosaiccanbeclas

3、sifiedintotwocategoriesaccordingtoareabasedandfeaturebasedaccordingtothebasicstepsofimagemosaicArea—basedmethodsdealwiththeimageswithoutattemptingtodetectfeaturesWindowsofpredefinedsizeorevenentireimagesareusedtomatchtwo

4、imagesThesemethodsalwaysspendmoretimeonestimation，andthemosaicresultwillbebadiftheoriginalimagesarenotgoodFeaturebasedmethodsconsistoffoursteps：featuredetection，featurematching，transformmodelestimationandimageresamplinga

5、ndtransformationTothesemethods，ifthecornersarenotdistinct，themosaicresuitalwayswillbenotgoodInthispaper，thetechnologyofimagemosaicwillbediscussed，andalsothetheoryofmultires01utionsignaldecompositionwillbeintroducedAndthe

6、n，anareabasedandafeaturebasedimagemosaicbasedonmultiresolutionsignaldecompositionwillbepresented，Inthearea—basedmethods，thetwosealewaveletWillbeused，andthewaveletcoefficientwillbemosaicbasedonareaAtlastreconstructingthem

7、osaiccoefficientthentheentireimagewillbegotInthefeature—basedmethods，thetwoscalewaveletalsowillbeusedtofindoutthefeaturepointsinthecoarsestlevelandthenthecorrelationofsensedimageandreferenceimagewillbefoundSOthetransform

8、ationmodelcanbeenestimatedtofinishthemosaicOfeachwaveletcoefficientLast，reconstructingthecoefficienttheresultwillbegotThesemethodsintroducedinthispaperwillimprovetheaccuracyofmosaicEspeciallyforthefeature—based，ifthecorn