資源描述:
《基于散亂數(shù)據(jù)的特征提取及曲面構(gòu)造技術(shù)研究》由會員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、華中科技大學(xué)碩士學(xué)位論文摘要反求工程是設(shè)計和加工復(fù)雜曲面零件的一種行之有效的手段,被廣泛地應(yīng)用于航天、航空、造船、汽車和模具制造業(yè)。為了解決反求工程中的大規(guī)模三維散亂數(shù)據(jù)的曲面重構(gòu)問題,本文對點(diǎn)云分割和曲面重建進(jìn)行了研究。//點(diǎn)云中不可避免的存在噪聲點(diǎn)。為了克服噪聲點(diǎn)的影響,選取較大的鄰域點(diǎn)集計算、點(diǎn)云特征,但較大的鄰域點(diǎn)集引入非當(dāng)前局部形狀的點(diǎn),從而導(dǎo)致局部特征計算的誤差,?<弋÷7一因此沐文提出基于修正曲率的特征計算方法,用較大的鄰域點(diǎn)集計算點(diǎn)云特征以克服噪聲點(diǎn)影響,同時根據(jù)點(diǎn)云分布對特征值進(jìn)行修正,從而突出邊界特征。散亂點(diǎn)云中相鄰點(diǎn)沒
2、有任何拓?fù)湫畔ⅲ瑸榱俗R別散亂點(diǎn)云的邊點(diǎn),本文提出了鄰域四邊形方法,將有規(guī)則的網(wǎng)格點(diǎn)云邊點(diǎn)識別方法應(yīng)用到散亂點(diǎn)云邊點(diǎn)識別中,并給出了鄰域四邊形的計算方法,從而準(zhǔn)確識別出散亂點(diǎn)云的邊點(diǎn)。本文對擬合曲線的點(diǎn)云進(jìn)行了排序、光滑、過濾處理,給出了多種處理方法,在實(shí)際中根據(jù)不同點(diǎn)云的特點(diǎn)交互選擇合適的方法:給出了截面點(diǎn)云的提取算法,以獲得反映點(diǎn)云截面形狀的截面曲線;完成了大規(guī)模點(diǎn)云的簡單曲面如平面、球、圓柱等的擬合。/本文以上述研究為基礎(chǔ),完成了交互式散亂數(shù)據(jù)的曲面重建模塊研制與開發(fā)工作,\并集成到三維造型軟件INTESOLID中。、p一一,,關(guān)鍵詞:
3、反求工程點(diǎn)云分割特征識別點(diǎn)云預(yù)處理曲面重建,,‘。—一—_●—一—~一+??一IABSTRACTReverseengineeringisanefficientmeasureforthedesignandmanufacturingofpartswithcomplexsurfaces,andithasbeenwidelyusedinaviation,shipbuilding,vehicleandmouldmanufacturingindustry.Tosolvetheproblemofsurfacereconstructionfromlarge
4、scalescattereddatainreverseengineering,therelevanttheoriesandpracticesofpointcloudsegmentationandsurfacereconstructionareinvestigated.Itisunavoidablethatthereexistnoisypointsinthepointcloud.Toavoidtheinfluencesofnoisypoints,abiggersetofneighborpointsareselectedforthecomput
5、ationofpointcloudfeature.However,thepointsthatdonotbelongtothelocalformswouldbealsointroduced,whichwouldleadtotheerrorofcomputation.Hence,thefemure—computingmethodbasedontherevisedcurvatureispresentedinthethesis,inwhichabiggersetofneighborpointsareselected,andthefeatureval
6、ueisrevisedbasedonthedistributionofpointcloudstohighlighttheboundaryfeatureofpointcloud.Thereexistsnotopologicalinformationbetweenneighborpointsinscatteredpointcloud.Toidentifytheedgepointsexactly,theneighborquadranglemethodispresented,whichappliestheidentificationmethodof
7、edgepointsinregulargridcloudstothescatteredpoints,andthecomputingmethodforneighborquadrangleisalsogiven.Hence,theedgepointscouldbeidentifiedfromthescatteredpointexactly.Severalmethodsarepresentedforthesorting,smoothingandfilteringofpointcloudsoffittingCUrVes,andtheappropri
8、atemethodisselectedinteractivelythroughanalyzingthepropeaiesofpointclouds.Thealgorithmofe