欢迎来到天天文库
浏览记录
ID:36357147
大小:3.26 MB
页数:105页
时间:2019-05-10
《数字视频修复若干关键技术研究——面向旧电影的视频修复和增强技术》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、分类号卫竖2lUDC1蜘87861密级——编号——中国科学院研究生院博士学位论文錾呈塑题堡复羞壬差毽燕盔珏塞=画包蝈电堂的塑题堡复垫塑堡撞查遘塞至指导教师玺谨盘熬援史国叠堂暄鱼王堂班窒压申请学位级别盟±学科专业名称一垡曼皇篮,皇处理论文提交日期至壁垒墨釜!旦!!目论文答辩日期2壁竖生£县签旦垮养单位主国型堂睦垒王堂婴塞逝学位授予单位主国型堂瞳盟究生暄答辩委员会主席耋堡塞熬援的运动连续性相结合起来,增加了检测算法的可靠性。5.基于已有的图像修复模型,将其扩展到三维,提出了一种基于空时结合图像修复技术的斑点去除法j该方法从数学形式上真正实现了空间信息和时间信息的相结合,充分利用了空间上和时间轴上
2、的相关信息,取碍了不错的斑点修复效果。关键词:数字视频修复,闪烁去除,斑点去除,总变分,图像修复AbstractFilmisaninvaluableartisticandculturalheritageinthelongtimeofhumanhistory.However,manyoldfilmsagebeinglostoragedbecausefilmisinafragileslateandonavolatilemedium.Itisextremelyneededtoconserveandrestoretheseoldfilms.Ontheotherhand,artisticcharman
3、dcommercialvalueoftheseoldfilmshaveattractedmoreand]/]orepeoplewhentherestoredfilmsareshowedtothepublicagain.Atthesametime,thesestoredoldfilmscanprovideabundantpro伊amsourcefordigital"IVandfilms.Withthedevelopmentofdigitaltechnology,digitalconservationandrestorationhasbecomegraduallythemainmethodfor
4、conservingandrestoringoldfilms.Digitalrestorationisatechnologyusingdigitalimageprocessingtorestoredegradationsexistinginoldfilmssuchasflicker,blotch,line·scratch,fadeandsoon.Todevelopasetofefficientandpracticalalgorithmsforcorrectingtheseartifactsisanattractivetopicofresearch.Nottoofewresearchresul
5、tsofdigitalrestorationhavebeenreportedintheoverseaswebs.NeverthelessnoresearchworkorreportcanbefoundinChina.Sotheresearchondigitalvideorestorationforoldfilmisconsideredasthemaincontentinthisdissertation,whereflkzkerremovalandblotchremovalaretheemphasesofresearch.Theinnovationsinthisdissertationcanh
6、elisledasfoIlows:1.Basedonthesmoothnesspropertyofflicker,amethoddetectinglocalmotionisproposedTheflickerparametersinthelocalmotionregionartdifferentfromthoseinthestillregion.Bymakinguseofthisfact,anapproximatelocalmotionregioncanbeflagged.Itishelpful'toestimateaccuratelyflickerparameter.2.Basedonsm
7、oothnessdifferencebetweentwosuccessiveframes,apartialdifferentialequationmethodforremovingintensityflickerisdeveloped.ThemethodUSetheideaofminimizationoftotalvariationof[inimage,Asaresull,themethodisnoise—t
此文档下载收益归作者所有