資源描述:
《圖象重建塊迭代算法的研究》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、北京交通大學(xué)碩士學(xué)位論文圖象重建塊迭代算法的研究姓名:孫杰申請學(xué)位級別:碩士專業(yè):計算數(shù)學(xué)指導(dǎo)教師:渠剛榮20071201』E巫窯適厶蘭亟±翌位絲塞空塞趟耍中文摘要摘要:在圖象重建算法中,最主要的兩種重建算法是解析重建算法和基于迭代的重建算法,Landweber迭代算法是圖象重建算法中基于迭代算法的重要圖象重建算法。本文將針對Landweber分塊迭代算法中松弛參數(shù)的選取進(jìn)行研究,在重建過程中采取對投影矩陣按投影角度分塊的方法,選取特定的松弛參數(shù)。按照本文采集數(shù)據(jù)模式進(jìn)行重建圖像。通過數(shù)值實驗得出結(jié)論:對于按角度分塊的塊迭代算法,松弛參數(shù)選取為^乘以塊矩陣與其
2、共軛轉(zhuǎn)置矩陣乘積的最大特征值分之~,當(dāng)采集完全投影數(shù)據(jù),且凡接近l/6-1/7時效果最好。另外,在本文中,按角度分塊的做法和松弛參數(shù)的選取方法對于有限角度圖象重建問題也是可行的,我們也做了相應(yīng)的實驗證實了這種可能性,并進(jìn)一步證明:當(dāng)圖像大小和每個投影角度下的射線條數(shù)一定時,投影角度增加,丑應(yīng)適當(dāng)變小,可以使得平均誤差最??;投影角度減小,A應(yīng)適當(dāng)變大,可以使得平均誤差最小。關(guān)鍵詞:代數(shù)重建算法:松弛參數(shù);塊矩陣:投影矩陣;有限角分類號:TP391.41』E夏至亟厶::j蘭亟上堂焦論塞ABSTRACT△墾S!基△£IABSTRACT:Thetwomostimportantimagereco
3、nstructionmethodsareanalyticalgorithmanditerativealgorithm.Landweber’Siterativealgorithmisanimportantmethodbasedontheiterativealgorithm.Inthisarticlewestudytherelaxationparameter'sselectionintheLandwebcr’Sblock—iterativealgorithms.Weselectthespecialrelaxationparanmterandputthepr
4、ojectionmatrixintomanyblocksaccordingtotheprojectionanglesinthereconstructionprocedure,andreconstructtheimageaccordingtotheformwegettheprojectiondatastrictlyWeselecttheproductofkandoneratiothemaximaleigenvaluesofthemultiplesoftheblock—matrixesandtheirconjugatetransposematrixesasthes
5、pecialrelaxationparameters.Byperformingnumericalexperiment,weconcludethatif入approachto116-1/7.wewillgetthebestresultwhenwecanobtainthecompletedatum.;Otherwise,thatweselectthespecialrelaxationparameterandputtheprojectionmatrixintomanyblocksaccordingtotheprojectionanglesisavailabletothelimite
6、danglepmblvm.Inthisarticle,wehavecertifiedthepossibilityoftheabovebyperformingtheconcretenumericalexperiment.Furthermore,wehavefoundedthatonconditionthatthesizeoftheimageandthenumberofthelinesateachofprojectionanglearecertain,wecanmaketheaverageerrorsmallerbyreducingthe旯properlywhe
7、nthenumberoftheprojectionanglesincreasegraduallyorincreasingtheAwhenthenumberoftheprojectionanglesdecrease.KEYWORDS:ART;Relaxationparameter;Block—matrix;Projectionmatrix;LimitedangleCLASSNo:11P391.41學(xué)位論文版權(quán)使