基于先驗信息的圖像恢復(fù)算法研究.pdf

1、分類號UDC基王盤墼籃盛兇圖堡迭復(fù)篡洼殛巒(中文題名)援毫晶(研究生姓名)密級Y85S473指導(dǎo)教師——盥L焦登職稱熬攫學(xué)位授予單位太連篷塞太堂申請學(xué)位級別～一蘧學(xué)科與專業(yè)籃呈皇籃息處理研究方向一一固遂信息墓統(tǒng)論文提交日期2嫂生3且學(xué)位授予日期答辯委員會主席』壹』繹評閱人生造一絲!二整AbstractClassicalalgorithmsofimagerestorationincludetheinversefilter，wienerfi

2、lter，constrainedleastsquaresfilterandSOonInthisthesis，blurimagerestorationmethodsbasedonpartialpriorinformationandconstrainedleastsquaresalgorithmbasedonorientationinformationmeasureareilluminatedindetailBlurimagerestora

3、tionbasedonthepartialpriorinformationhasbeendevelopedinthispaper，andgetsPSFusingaruleofcorrespondencebetweenthepartialpriorinformationwiththeblurredimageFirst，afinitesupportimageofpriorinformationisgotSecond，afinitesuppo

4、rtblurimagematchingtopartialpriorinformationisobtainedwithimagematchingalgorithmbasedoncorrelationcoefficientThird，trueblurimagecorrespondingtopriorinformationisacquiredwithiterativemethodofenumerationFourth，prioriblurid

5、entificationgetsPSFbyWienerFiltermethodFinally，entireimageisestimatedwithwienerfiltermethodThispaperputsforwardconstrainedleastsquaresalgorithmbasedontheorientationinformationmeasuretoo，Wecreateedgerestrictionusingedge—p

6、reservingpropertyoforientationinformationmeasure，edge—informationisconsideredtoanovelconstrainedcondition，andparameterofnoiseandedge—informationisestimatedbygeneratedcross’validation(GCV)methodNoiseissmoothedandedgeinfor

7、mationispreservedinresultoftheimagerestoration，itisbetteradaptedtohuman’SvisualcharacteristicOntheotherhand，severalideasandfutureworkarewrittenintheendofthepaperbasedontheconclusionofforenamedmethodsKeyWords：hnageRestora