面向超分辨率重建的圖像配準(zhǔn)方法研究.pdf

1、大連海事大學(xué)碩士學(xué)位論文面向超分辨率重建的圖像配準(zhǔn)方法研究姓名：張燕申請學(xué)位級別：碩士專業(yè)：信號與信息處理指導(dǎo)教師：謝公福王新年20080301英文摘要ResearchonSuperresolutionReconstructionOrientedImageRegistrationAbstractWiththedevelopmentofmultimediatechnique，muchhigherdefinitioniSneededtosa

2、tisfytheincreasingdemandSuperresolutionrestorationofthevideosequenceisthedigitalsignalprocessingmethod，whichisusedtoimprovetheimageresolutionItisalsoakindofimageenhancementtechniquewhichisusedtoconstructoneormoresuperres

3、olutionimagesfromthedegradedandlowresolutionobservedsequenceItisrelatedtoimageregistration，imagerestoration，informationrenderingandfusiontechniqueImagerestorationisusedtorestore黟ayscaleresolutionfromtheobserveddegradedim

4、age；imagefusionistocombineinformationfrommultipleimagestomakethetargetimagemuchclearerImageregistrationwhichisdefinedasmappingbetweentwoimagesbothspatiallyandwithrespecttointensityItsaccuracydeterminestheperformanceofsup

5、erresolutionAtpresent，themainproblemsoftheexistingimageregistrationalgorithmsappliedtosuperresolutionare“(1)Mostofsuperresolutionmethodsdon’tpayenoughattentiontoimageregistration；(2)Theexistingimageregistrationalgorithms

6、arealmostbasedontheassumptionthatasidefromthetranslationcomponents，theotherparametersvaryverysmallinthesequenceInmanypracticalsituations，howevertheseassumptionsalealmostnevermetTheaccuracyofimageregistrationresultsCan’ts

7、atisfytherequirementsofsuperresolutionwhenthesequenceimages’rotationangleandthescalefactorbetweentheimagestoberegisteredchangeinalargescaleThisthesis，accordingtotheproblemsmentionedabove，proposedthealgorithmofthesequence

8、imageregistrationbasedontargettrackingFirstlytheinterestedtargetareaisroughlylocatedbyusingthemethodbased011SIFTtargettracking，thenthemotionparametersarecalculated，andfinallytheseparametersarerefinedbyusingthemethodofLuc