d. improved image quilting

d. improved image quilting

ID:34568747

大小:1.45 MB

页数:8页

时间:2019-03-08

d. improved image quilting_第1页
d. improved image quilting_第2页
d. improved image quilting_第3页
d. improved image quilting_第4页
d. improved image quilting_第5页
资源描述:

《d. improved image quilting》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库

1、ImprovedImageQuiltingJeremyLongDavidMouldDepartmentofComputerScienceDepartmentofComputerScience.UniversityofSaskatchewanUniversityofSaskatchewanjsl847@mail.usask.camould@cs.usask.caFigure1:ImageQuilting(left)andImprovedImageQuilting(right).ABSTRACTsearchdoneinthisarea,wefocusveryspecificallyonthed

2、omainwhereourcontributionismostrelevant.Assuch,ouremphasisInthispaper,wepresentanimprovementtotheminimumer-isonpatch-basedsynthesisandonpreviousmethodsforavoidingrorboundarycut,amethodofshapingtexturepatchesfornon-boundaryartifactsalongtheseamsofthepatches.Wealsosurveyparametrictexturesynthesisfr

3、omexamplealgorithmssuchasEfrosseveralotherdomainswhereimprovementstotheminimumerrorandFreeman’sImageQuilting[4].Ourmethodusesanalternateboundarycutmightprovefruitful,suchasWangTiles[13,14].distancemetricforDijkstra’salgorithm[3],andasaresultweareOncewehaveabackgroundforcomparison,wethendescribeab

4、letopreventthepathfromtakingshortcutsthroughhighcostthealternatenon-scalardistancemetricthatourimprovedimageareas,ascansometimesbeseenintraditionalimagequilting.Fur-quiltingalgorithmuses.Thismetricencouragespathstoavoidpeakthermore,ourmethodisabletoreduceboththemaximumerrorincostareas,evenifitmea

5、nstakinglongcircuitousroutes.Weshowtheresultingtextureandthevisibilityoftheremainingdefectsbyinthispaperhowthispropertyisdesirablefortexturesynthesisspreadingthemoveralongerpath.Post-processmethodssuchasandotherapplicationswheretheminimumerrorboundarycutispixelre-synthesis[9]caneasilybemodifiedand

6、appliedtoourapplied.Wealsoshowhowthismetriccanbeapproximatedtofitminimumboundarycuttoincreasethequalityoftheresults.withauser’spreferences.Finally,wecompareresultsobtainedusingseveraldifferentKeywords:ACMclassification:1.3.7:Three-Dimensionalmethodsforshapingthetexturepatches.WefindthatourmethodGrap

7、hicaandRealism,1.4.7FeatureMeasurement,texture,non-tendstoavoidthemostgrievousdiscontinuities,bothbecauseitparametrictexturesynthesisfromexample,minimumerrorbound-activelydecreasesthemaximumerroralongtheboundarycutanda

当前文档最多预览五页,下载文档查看全文

此文档下载收益归作者所有

当前文档最多预览五页,下载文档查看全文
温馨提示:
1. 部分包含数学公式或PPT动画的文件,查看预览时可能会显示错乱或异常,文件下载后无此问题,请放心下载。
2. 本文档由用户上传,版权归属用户,天天文库负责整理代发布。如果您对本文档版权有争议请及时联系客服。
3. 下载前请仔细阅读文档内容,确认文档内容符合您的需求后进行下载,若出现内容与标题不符可向本站投诉处理。
4. 下载文档时可能由于网络波动等原因无法下载或下载错误,付费完成后未能成功下载的用户请联系客服处理。