資源描述:
《單幅霧天圖像的復(fù)原和增強(qiáng)算法的研究》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、摘要霧是由于空氣中漂浮的水汽凝結(jié)而造成能見度降低的一種大氣現(xiàn)像。在霧天條件下,因?yàn)槟芤姸鹊南陆?,從而?dǎo)致室外圖像采集系統(tǒng)獲得的圖像的清晰度下降,致使圖像退化而變得模糊,這極大的影響圖像信息的提取,并且極大地限制和影響了室外視覺系統(tǒng)(如室外的交通檢察、導(dǎo)航系統(tǒng)、物體追蹤等)的正常工作,也對(duì)人們的出行產(chǎn)生了極大地影響。因此對(duì)因霧天情況造成的圖像品質(zhì)下降的問題開展研究具有普遍的實(shí)際意義。當(dāng)前,對(duì)于霧天圖像的清晰化算法主要有兩大方向。一個(gè)方向?yàn)閳D像增強(qiáng)的方法。因?yàn)橥嘶蟮膱D像的亮度和對(duì)比度都相對(duì)較低,圖像增強(qiáng)的方法只從圖像的亮度和對(duì)比度的角度來起到圖像增強(qiáng)的作用,
2、它不考慮圖像退化的原因。目前使用較多的是直方圖均衡化和基于Retinex理論的MSR算法。另一種是圖像復(fù)原的方法。這種方法一般要分析圖像的退化原因,通過退化原因來建立霧天圖像的退化模型,利用霧天圖像的退化模型對(duì)圖像進(jìn)行圖像復(fù)原。本文通過優(yōu)化大氣物理退化模型,在多尺度Retinex算法(Multi-ScaleRetinex,MSR)的基礎(chǔ)上,提出了一種新的去霧方法。該方法能有效去除霧化效果,并且失真率比較小,實(shí)現(xiàn)彩色霧天退化圖像的復(fù)原。關(guān)鍵詞:去霧,圖像增強(qiáng),圖像復(fù)原,多尺度Retinex算法IIAbstractHazearesuchphenomenawhi
3、charousedbyatmosphericabsorptionofscatteredbecauseofparticlesandwater-dropletsintheatmosphere.Inhazeday,theimagesobtainedbytheopticalequipmenthavedecreasedcontrastbecauseoftheatmosphericvisibilitydecreased,aswellasimagesdegradationandblur,thusaffectingtheimageinformationextraction
4、.Thisnotonlygreatlylimitandinfluencesthenormaloperateofoutdoorvisionsystems(suchasoutdoormonitoring,automaticnavigation,targettracking,etc.),butalsoseriouslyaffectpeople'sdailylives.Sostudyingonimagequalitydecreasedcausebythenaturalphenomenonhaspracticalsignificance.Recently,image
5、foggyremovalshavetwomajortechniques:oneisimageenhancement,whichneedsonlyconsiderthecharacteristicsofthelowbrightnessandlowcontrast,doesnotconsiderthereasonsofimagedegradation.Wecancallittheimageenhancement.Theothertechniqueiscreatingamodefromtheperspectiveofthespecificandphysicalc
6、ausesoftheimagesdegradation,andthenresumethesceneusedthemodel.Inotherwords,itmodelsthedegradationprocess.Thismethodisusuallyreferredtoimagerestoration.BasedonMulti-ScaleRetinex(MSR)algorithm,anewmethodcombinedwithatmosphericphysicsmodelwasproposedinthispaper.Theexperimentalresults
7、demonstratethesatisfyingvisualeffectsofthisapproach.Keywords:Defogging;imageenhancement;imagerestoration;Multi-scaleRetinexIII目錄摘要..........................................................................................................................................IABSTRACT....
8、.................................