視頻編碼中的運(yùn)動(dòng)估計(jì)技術(shù)研究.pdf

1、分類號(hào)!№!UDC密級(jí)編號(hào)中國(guó)科學(xué)院研究生院碩士學(xué)位論文Y8【J哇乏21奎至堡指導(dǎo)教師掛渲歪副嬰窺雖蟲國(guó)型堂瞳自動(dòng)絲盟塞逝申請(qǐng)學(xué)位級(jí)別亟學(xué)科專業(yè)名稱讓簋扭廑旦遮盔論文提交日期20055論文答辯日期20056培養(yǎng)單位空墾型堂暄自動(dòng)絲硒塞壓學(xué)位授予單位史圍型堂瞳塑究生院答辯委員會(huì)主席中國(guó)科學(xué)院自動(dòng)化研究所碩士學(xué)位論文Abs訂actAbstlactlnthevideocodmgandprocessingsystem，motionestimat

2、ion(ME)andⅡlotioncompensation(MC)playaveryimpon趾tmleTheycaIlelimin砒etheinter丹拋eredundallcyandiIllpmVe也eperfbrnlanceofthe“deocodeLOn也eonehalld，meaccuracyofmotionestima廿ondirectlya能ctstheemciencyofthevideocodeLIteliminates

3、ⅡleiIlte卜fhmecorrelationandreducesthedemaIldofthestorememoryandnlebandwid也fortransmitting掣eauyOntheo也erha芏ld，mernotionestimationcostsmostofthecodingtimeof也eencoder，thereforetllecomplexityoftheencoderliesonthecomplex時(shí)ofth

4、emotionestimationalgorimmIn廿listhesis，wemainlydiscussedmemotionestima垃onandmotioncompensationtechlliquesMoredetailedcontentof也ist11esiscanbedescribedasf01lows：FirsUy，thesub—pixelmotionestimationteclllliquesaredeeplystlld

5、iedThcsub—pixeImotionestimationcanbedividedjntot、陽(yáng)st印s：sub巾ixe】jn柵polationaIldsubpixelmatcllingInsub—pixel缸elpol撕on，廿1eapplicationsofmuItitap6lterandtheef琵ctofthetapsoftlle矗1tertot】1epfecisjonandcomplexityoftllesublpixe】

6、interpolationalgorimmarestudiedInsubpixelmatching，afastqu礎(chǔ)er_pixelmotionestimationa190ritlllnisproposed，basedonthesta缸sticalallalysjsofthedistributionofquane卜pixelprecisionmotionvectorsThealgori血mreducesthecomplexityofqu

7、arter_pixelmotionestimationgreatlySecondly’memllltiblockmodemotionest豳眥ionalgorithmofverylowbitm￡evideosequencesisstI工diedTheinter_舳memotionoftheverylowbit_mtevideosequencesismuchlesscompared、vithtllatofnlehi曲bit—ratevid

8、eosequences，thusthebitrateis瑚LuchloweLT置1rou曲analyzingthecharacteristicofmeverylowbit—ratevideosequencesandtb【ecorrclationoftllemIlltiblockmodes，afastquarterpixelmotionestimationaIgorithmisproposed，whicho丘brst11ecomputat

9、ionalconlpIexi母andperfornlancetrade—offsinaVerylowbitrateVideocodingschemeFinally，themul廿blockmodemotionestimationisstudiedBasedonmestatisticalcharacteristicof【11emodedismb謝on，twomulti_blockmodemotionestimationalgori廿lms