method

1AssessmentSheetMethod15%PAPERINFORMATIONTitleofthePaper(1)3Dmotionestimationfordepthimagecodingin3Dvideocoding(2)Globally-OptimalGreedyAlgorithmsforTrackingaVariableNumberofObjects(3)Imagesegmentationwithsimultaneousilluminationandreflectanceestimation:anenergyminimizationapproachNumberofPages(1)7pages;(2)8pages；(3)7pages；Subject学科Ⅰ.LANGUAGE5%MARK:______1.LexicalFeaturesHighly-TechnicalWordsmotioncompensation(运动补偿),temporalredundancy（时间冗余）,motionvector（运动矢量）,bitrate（比特率）,pixel（像素）,referenceframe（参考帧）,coder（编码器）,mean-squarederror（均方误差）,meanabsoluteerror（绝对平均误差）,depthmap（深度图）,sub-pixelprecision（亚像素精度）,bitstream（比特流）(1)multi-objecttracking（多目标跟踪）,non-maxsuppression（非极大值抑制）,maximuimaposteriori（最大后验）,graph-cut（图割）(2)illuminationimage（照度图像）,imagesegmentation（图像分割）,graylevelimage（灰度图）,Lagrangemultiplier（拉格朗日乘子）(3)NewlyCoinedWordsPosteriori(2)Non-technicalWords（formal）:perform,obtain,successive,technique,implement,reasonable,accurate,partition,suitable,evaluate,represent,compute,approach,optimal,dimension,attempt,horizontal,vertical,improve,indicate,correspond,limit,vary,minimize,define,denote,upper,select,desirable,address,correlation,observe,assumption,share,achieve,exploit,construct(1)equivalent,derive,perspective,consider,location,specify,generate,transition,state,behave,collection,tend,model,border,score,describe,likelihood,imply,foreground,occlude,cluster,procedure,embed,drop,preserve,split,merge,span,produce,iterate,feasible,sufficient,modify,cache,operate(2)additive,extend,specific,property,smooth,express,seek,task,distribution,simultaneous,approximate,intensity,review,determine,notation,regularize,quadratic,confirm,remove(3)Semi-TechnicalWordsframe,residual,match,distortion,map,error,function,mean,coordinate,sequence,difference(1)variable,set,weight,order(2)constant,term(3)Useofnounphraseswithexpandedpreandpost-modificationvideocodingtechniques,blockbasedmotioncompensation,mean-squarederror,3Dblockmatchingalgorithm,three-dimensionalmotionvectors,searchingwindow,matchingfunction,correspondingblock,codingefficiency(1)vector-valuedrandomvariable,sliding-windowobjectdetector,explicitnon-overlappingconstraint,totalrunningtime,multi-objecttrackingproblem,dynamicprogrammingalgorithms,globally-optimalalgorithm,negative-costdirectedcycle,forward-progressingpath(2)2theoreticalcompleteness,piecewiseconstantmap,correspondingreflectanceconstants,obtainedmembershipfunctions,energyminimizationframework,observedimageintensities,clustering-basedapproach,basicclusteringcriterion,standardclusteringalgorithm,clusteringcriterionfunction,spatiallyvaryingilluminationimagecomponent,derivedoptimalilluminationimage(3)Useofprepositionalphrasesandparticiplephrasesasadverbialorrestrictivephraseswithinthescenerelatingto,shiftingdirectionofblocks,basedonthespeedofobjects,withmotionsharing,insteadofencodingmotionvectors,representedbythree-dimensionalmotionvectors,performedoverthereferenceframe,definedasfollows,matchingindepthdirection,predictedfromneighboringblocks(1)fortheprobabilityofatrack,considerablereductionincomplexity,transitioningintoaterminationstate,observedatallspace-timelocations,scoredbyasliding-windowobjectdetector,generatedfromaforegroundappearancemodel,withaslightabuseofnotation,givenbyourdynamicmodel(2)variationinappearance,alongwiththepiecewiseconstantproperty,inviewoftheaboveclusteringcharacterization,expressedasalinearcombination,achievedbyseekingmembershipfunctions,associatedwiththeregions,exploitedintheproposedmethod,appliedforlocalintensityclassification(3)UseofnominalizationThentheblockmatchingisperformedinthedepthdirectiontofindthezmotionvectorcomponentwhichminimizesthemotioncompensationerror(1)……theshiftingdirectionofblocksinthereferenceframe……(1)Bysuppressingextradetectionsaroundeachtrackasitisinstanced,weallowforthepossibilitythatthepriorcanoverridetheobservationtermandselectawindowwhichisnotalocalmaxima.(2)Therecentworkof[2]makeasimilarargumentandaddanexplicitnon-overlappingconstrainttotheirILP,whichmaysacrificetractability.(2)Thevariationinappearancecausedbytheilluminationchangeshasbeenachallengingproblemforvisualtasks,suchassegmentation.(3)Forimageswithhighlevelnoise,itisnecessarytoregularizemembershipfunctionbyaddingaregularizationtermR(U)totheaboveclusteringenergyJ(U,r,S).(3)Verbs(materialpreparation,“equipment”placement,procedureetc):perform,obtain,vary,minimize,select,share,reduce,achieve,predict,construct,provide,encode,partition,evaluate(1)initialize,iterate,find,update,increase,decrease,terminate,start,yield,modify,introduce,compute,convert,require,run,return,cache,reconstruct,operate,mark,label,share,propagate,recomputed(2)extend,express,approximate,perform,minimize,initialize,apply,obtain(3)2.SyntacticFeaturesSentencepatterns(typicalsentenceexamples)TenseVoiceBeginningof“M”sectionInvideocoding,motioncompensationisusedforreducingthetemporalredundancybetweensuccessiveframesofavideosignal.(1)Wedefineanobjectivefunctionformulti-objecttrackingequivalenttothatof[25].(2)Fortheoreticalcompleteness,weslig