基于幀聚類和譜小波的網(wǎng)格動(dòng)畫壓縮算法.pdf

1、摘要基于幀聚類和譜小波的網(wǎng)格動(dòng)畫壓縮算法摘要近年來，三維網(wǎng)格動(dòng)畫被廣泛應(yīng)用于在線三維游戲、遠(yuǎn)程醫(yī)療系統(tǒng)等不同領(lǐng)域。為了表現(xiàn)更強(qiáng)烈的真實(shí)感，三維網(wǎng)格動(dòng)畫的細(xì)節(jié)不斷豐富，從而導(dǎo)致網(wǎng)格動(dòng)畫的數(shù)據(jù)量也呈指數(shù)倍擴(kuò)大。因此設(shè)計(jì)網(wǎng)格動(dòng)畫數(shù)據(jù)的壓縮算法以減少傳輸帶寬和降低存儲(chǔ)空間，對(duì)網(wǎng)格動(dòng)畫的發(fā)展及應(yīng)用具有十分重要的意義。為此，本文對(duì)三維網(wǎng)格序列動(dòng)畫壓縮進(jìn)行了深入的研究，主要完成了以下幾方面的工作。首先，提出一種可用于動(dòng)畫漸進(jìn)傳輸?shù)膸垲愃惴?。該算法?/p>

2、定義幀相似度時(shí)以網(wǎng)格殘差替代傳統(tǒng)的歐氏距離，然后再加入關(guān)鍵幀時(shí)間差值，使得聚類結(jié)果不僅準(zhǔn)確，還能保證各類所包含的幀索引連續(xù)，從而符合動(dòng)畫漸進(jìn)傳輸?shù)囊?。其次，提出兩種網(wǎng)格信號(hào)定義方式，為下一步譜小波變換做準(zhǔn)備。主要包括：1)將各類代表幀與類內(nèi)剩余幀的網(wǎng)格殘差作為一種信號(hào)；2)提出一種改進(jìn)的軌跡PCA降維算法，分別在x，y，z方向構(gòu)建軌跡矩陣后執(zhí)行PCA降維，然后將得到的PCA系數(shù)作為一種網(wǎng)格信號(hào)。再次，提出一種用于網(wǎng)格壓縮的譜圖小波變換

3、?；谧V圖理論T碩士學(xué)位論文ABSTRACTMESHANIMATIoNCoMPRESSIONVIAFRAMECLUSTERINGANDSPECTRALGRAPHABSTRACTI心Y帆2㈣75咖叭1帆0㈣68帆I(xiàn)nrecentyears，3Dmeshanimationisappliedondifferentfieldssuchas3DonlinegamesandremotemedicalsystemInordertoshowstrong

4、realisticsense，enrichingthedetailsof3DmeshanimationconstantlywillbringexponentialincreasingdataTherefore，ithasagreatsignificanceonthedevelopmentandapplicationofmeshanimationteducethenetworkbandwidthandstoragespacethrough

5、designingvariousnewanimationcompressionalgorithmsThecompressionof3Dmeshanimationisstudiedsystematicallyinthispaper，andthemainworkiscompletedasfollowsFirstlyakindofframe—clusteringalgorithmwhichcanbeusedforprogressivetran

6、smissionofmeshanimationisproposedinthispaperThispapernotonlyusesthemeshresidualsteplacetheoriginalEuclideandistance，butalsoaddthekey—timedifferenceasapartofflame—distanceBythat，wecanensuretheindexofframesineachclusterkee

7、pcontinuousSecondlythispapertryingtousetwowaystogetsignalswhichdefinedonmeshesforspectrumwaveletcompression：1)Selecttherepresentativeframesfromeachclusterandthencomputethemeshresidualsbetweentherepresentativemeshandtheot