简介:为了提高图像插值的恢复效果,提出了一种基于图结构正则化稀疏表示的双层伯格曼迭代算法.该迭代算法的外层用于约束图像观测数据,内层用于更新图像块的学习字典和稀疏表示系数.引入的图结构正则化稀疏表示约束可以有效地自适应图像块的局部结构,对于严重受损的情形也能得到精确的恢复结果.此外,在内层迭代中改进的稀疏表示和简洁的字典更新策略使算法能快速地趋于收敛.数值实验结果表明,所提出的算法可以有效地恢复图像,在主观视觉效果和客观量化标准上要优于目前已有的算法.
简介:为了提高图像插值的恢复效果,提出了一种基于图结构正则化稀疏表示的双层伯格曼迭代算法.该迭代算法的外层用于约束图像观测数据,内层用于更新图像块的学习字典和稀疏表示系数.引入的图结构正则化稀疏表示约束可以有效地自适应图像块的局部结构,对于严重受损的情形也能得到精确的恢复结果.此外,在内层迭代中改进的稀疏表示和简洁的字典更新策略使算法能快速地趋于收敛.数值实验结果表明,所提出的算法可以有效地恢复图像,在主观视觉效果和客观量化标准上要优于目前已有的算法.更多还原
简介:利用Jensen不等式,Steklov变换,Cauchy积分主值讨论了一类离散指数型线性积分修正插值算子在Orlicz空间L*M(-∞,∞)中的逼近问题,给出了收敛速度的估计.
简介:LedbyfourgenerationsofleadershipfromlateProf.JIANGSichang(academician,ChineseAcademyofEngineering),Prof.YANGWeiyan(HonoraryPresident,DivisionofOtolaryngologyHeadandNeckSurgery,ChineseMedicalAssociation),Prof.HANDongyi(PresidentElected,DivisionofOtolaryngologyHeadandNeckSurgery,ChineseMedicalAssociation)tonowProf.YANGShiming(President,DivisionofOtolaryngologists,
简介:Lightweightdesignisoneofthedevelopmenttrendsoftheautomobileindustry.AneffectivewaytoachievelightweightautobodiesistouseAHSS(advancedhighstrengthsteel)forthesafetycomponentsofautomobiles.ThisstudyhastakendoorsillreinforcementsmadeofmartensiteAHSSastheobject,andperformedresearchontheAHSSrollformingtechnologiesandprototypedevelopmentoftypicalasymmetricopencomponents.Bymeansoffiniteelementanalysis(FEA)andsimulation,studieshavebeencarriedoutonthespringbackandedgewavedefectsduringAHSSrollforming,andanoptimizedprocessdesignhasbeenachieved.Thegenerationmechanismsofverticalbows,horizontalcambers,twists,pre-punchedholedistortionandcutendflarehavebeenanalyzed,andsolutionstothesedefectshavebeengiven.Inaddition,tesingoftherollformingprocessforAHSShasbeenconductedandtypicalsampleswithrequireddimensionalaccuracyhavebeenmanufactured.Thisstudyhasprovidedtechnicalsupportforthelarge-scaleapplicationofAHSS.
简介:Large-scaleMIMO(multiple-inputmultiple-output)systemswithnumerouslow-powerantennascanprovidebetterperformanceintermsofspectrumefficiency,powersavingandlinkreliabilitythanconventionalMIMO.Forlarge-scaleMIMO,thereareseveraltechnicalissuesthatneedtobepracticallyaddressed(e.g.,pilotpatterndesignandlow-powertransmissiondesign)andtheoreticallyaddressed(e.g.,capacitybound,channelestimation,andpowerallocationstrategies).Inthispaper,weanalyzethesumrateupperboundoflarge-scaleMIMO,investigateitskeytechnologiesincludingchannelestimation,downlinkprecoding,anduplinkdetection.Wealsopresentsomeperspectivesconcerningnewchannelmodelingapproaches,advanceduserschedulingalgorithms,etc.
简介:Opticalmotioncaptureisanincreasinglypopularanimationtechnique.Inthelastfewyears,plentyofmethodshavebeenproposedforkey-frameextractionofmotioncapturedata,anditisacommonmethodtoextractkey-frameusingquaternion.Here,onemaindifficultyisduetothefactthatpreviousalgorithmsoftenneedtomanuallysetvariousparameters.Inaddition,itisproblematictopredefinetheappropriatethresholdwithoutknowingthedatacontent.Inthispaper,wepresentanoveladaptivethreshold-basedextractionmethod.Key-framecanbefoundaccordingtoquaterniondistance.Weproposeasimpleandefficientalgorithmtoextractkey-framefromamotionsequencebasedonadaptivethreshold.Itisconvenientwithnoneedtopredefineparameterstomeetcertaincompressionratio.Experimentalresultsofmanymotioncaptureswithdifferenttraitsdemonstrategoodperformanceoftheproposedalgorithm.Ourexperimentsshowthatonecantypicallycutdowntheprocessofextractionfromseveralminutestoacoupleofseconds.