Single Image Haze Removal Using Dark Channel P...

Single Image Haze Removal Using Dark Channel P...

ID:40101723

大小:2.45 MB

页数:8页

时间:2019-07-21

Single Image Haze Removal Using Dark Channel P..._第1页
Single Image Haze Removal Using Dark Channel P..._第2页
Single Image Haze Removal Using Dark Channel P..._第3页
Single Image Haze Removal Using Dark Channel P..._第4页
Single Image Haze Removal Using Dark Channel P..._第5页
资源描述:

《Single Image Haze Removal Using Dark Channel P...》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库

1、SingleImageHazeRemovalUsingDarkChannelPriorKaimingHe1JianSun2XiaoouTang1,21TheChineseUniversityofHongKong2MicrosoftResearchAsiaAbstractInthispaper,weproposeasimplebuteffectiveimageprior-darkchannelpriortoremovehazefromasin-gleinputimage.Thedarkchannelpriorisakindofstatis-ticsofthehaze-f

2、reeoutdoorimages.Itisbasedonakeyobservation-mostlocalpatchesinhaze-freeoutdoorim-agescontainsomepixelswhichhaveverylowintensitiesinatleastonecolorchannel.Usingthispriorwiththehazeimagingmodel,wecandirectlyestimatethethicknessofthehazeandrecoverahighqualityhaze-freeimage.Resultsonavariet

3、yofoutdoorhazeimagesdemonstratethepoweroftheproposedprior.Moreover,ahighqualitydepthmapcanalsobeobtainedasaby-productofhazeremoval.Figure1.Hazeremovalusingasingleimage.(a)inputhazeimage.(b)imageafterhazeremovalbyourapproach.(c)ourrecovered1.Introductiondepthmap.Imagesofoutdoorscenesareu

4、suallydegradedbytheturbidmedium(e.g.,particles,water-droplets)intheatmo-contrastsceneradiance.Last,thehazeremovalcanproducesphere.Haze,fog,andsmokearesuchphenomenaduedepthinformationandbenefitmanyvisionalgorithmsandtoatmosphericabsorptionandscattering.Theirradianceadvancedimageediting.Ha

5、zeorfogcanbeausefuldepthreceivedbythecamerafromthescenepointisattenuatedclueforsceneunderstanding.Thebadhazeimagecanbealongthelineofsight.Furthermore,theincominglightisputtogooduse.blendedwiththeairlight[6](ambientlightreflectedintoHowever,hazeremovalisachallengingproblembecausethelineof

6、sightbyatmosphericparticles).Thedegradedthehazeisdependentontheunknowndepthinformation.imageslosethecontrastandcolorfidelity,asshowninFig-Theproblemisunder-constrainediftheinputisonlyasin-ure1(a).Sincetheamountofscatteringdependsonthedis-glehazeimage.Therefore,manymethodshavebeenpro-tanc

7、esofthescenepointsfromthecamera,thedegradationposedbyusingmultipleimagesoradditionalinformation.isspatial-variant.Polarizationbasedmethods[14,15]removethehazeeffectHazeremoval1(ordehazing)ishighlydesiredinboththroughtwoormoreimagestakenwithdifferentdegreesofconsumer/computation

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。