資源描述:
《基于各向異性濾波和空間FCM的MRI圖像分割方法-論文.pdf》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、第31卷第1期計算機應(yīng)用研究V01.31No.12014年1月ApplicationResearchofComputersJan.2014基于各向異性濾波和空間FCM的Mill圖像分割方法木曾文權(quán),何擁軍,崔曉坤(廣東科學技術(shù)職業(yè)學院計算機工程技術(shù)學院,廣東珠海519090)摘要:針對具復(fù)雜目標和邊界模糊的MRI圖像中多感興趣區(qū)域的分割中分割MRI圖像軟組織難的問題,提出了一種基于各向異性濾波和空間模糊C一均值聚類(SFCM)的MRI圖像分割方法;用新型各向異性濾波對圖像進行預(yù)處理,解決去噪平滑的同時弱化圖像細節(jié)的問題;用鄰域空間信息設(shè)計空間函數(shù),改進傳統(tǒng)FCM的目標函數(shù)
2、:用圖像的空間信息實現(xiàn)圖像各目標準確分類、有效解決孤立區(qū)域的正確歸類問題,進而使分割區(qū)域完整;用直方圖擬合曲線初始化分類數(shù)和初始聚類中心,加快算法迭代到最優(yōu)解,進而減少運行時間。通過實驗證實了各向異性濾波和空間FCM的MRI圖像分割方法的綜合應(yīng)用顯著提高了分割灰度重疊、目標不連續(xù)和目標邊界模糊的MRI圖像的分割效果。關(guān)鍵詞:磁共振成像;圖像分割;各向異性擴散;FCM;空間FCM中圖分類號:TP391文獻標志碼:A文章編號:1001—3695(2014)01-0316一O5doi:10.3969/j.issn.1001—3695.2014.01.075MRIimageseg
3、mentationmethodbasedonanisotropicdiffusionandspatialFCMZENGWen—quan,HEYong-jun,CUIXiao—kun(CollegeofComputerEngineeringTechnical,GuangdongInstituteofScience&Technology,ZhuhaiGuangdong519090,China)Abstract:T0resolvethedifficultproblemofsofttissuesegmentationofMRIimagesofthesegmentationofth
4、emuhitargetre—gionofinterestintheMRIimageswithcomplextargetsandfuzzyboundary,thispaperproposedanovelMRIimagesegmenta—tionmethod.basedonanisotropicdiffusionandspatialfuzzyC.meansclustering(SFCM).Itpreprocessedtheimagesusingthenonlinearandanisotropicdiflusion,resolvingtheproblemofweakeningt
5、heimagedetailswhileremovingthenoise.Itdesignedspacefunctioncombiningwiththeneighborhoodspace.improvingtraditionalFCMobjectivefunction.Itusedthespatialinfor—mationoftheimagetoachievetheaccurateclassificationofeveryobjeetinimagewasaneffectivesolutiontotheisolatedareacorrectlyclassified.Afte
6、rthat.jtobtainedcompleteandcontinuoussegmentedregions.Finally,itutilizedthefittingcurveofhistogramtoinitializetheclassificationnumberandtheinitialclustercenters.a(chǎn)cceleratingthealgorithmiterativetotheoptimalsolution.a(chǎn)ndalsoreducingtheruntime.Theexperimentsshowthatthepossibilityoffindabests
7、olutioniSimprovedbyintro—ducingthemethodofMRIimagesegmentationmethodbasedonanisotropicdiffusionandspatialFCM.soastotheprocessingofMRIimagessegmentationwhichhasoverlappedgrayscale,discontinuousobjectsandfuzzyboundary.KeyWOrds:MRI:imagessegmentation;anisotropicdiffusi