2010-Construction of a Large Class of Deterministic

2010-Construction of a Large Class of Deterministic

ID:39887766

大小:276.33 KB

页数:16页

时间:2019-07-14

2010-Construction of a Large Class of Deterministic_第1页
2010-Construction of a Large Class of Deterministic_第2页
2010-Construction of a Large Class of Deterministic_第3页
2010-Construction of a Large Class of Deterministic_第4页
2010-Construction of a Large Class of Deterministic_第5页
资源描述:

《2010-Construction of a Large Class of Deterministic》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、1ConstructionofaLargeClassofDeterministicSensingMatricesthatSatisfyaStatisticalIsometryPropertyRobertCalderbank,Fellow,IEEE,StephenHoward,Member,IEEE,andSinaJafarpour,StudentMember,IEEEAbstract—CompressedSensingaimstocaptureattributesofThetwofundamentalquestionsincompressedsensing

2、are:k-sparsesignalsusingveryfewmeasurements.InthestandardhowtoconstructsuitablesensingmatricesΦ,andhowtoCompressedSensingparadigm,theN×Cmeasurementmatrixrecoverαfromfefficiently;itisalsoofpracticalimportanceΦisrequiredtoactasanearisometryonthesetofalltoberesilienttomeasurementnoise

3、andtobeabletorecon-k-sparsesignals(RestrictedIsometryPropertyorRIP).IfΦsatisfiestheRIP,thenBasisPursuitorMatchingPursuitrecoverystruct(approximationsto)k-compressiblesignals,i.e.signalsalgorithmscanbeusedtorecoveranyk-sparsevectorαfromthathavemorethanknonvanishingentries,butwhereon

4、lyktheNmeasurementsΦα.AlthoughitisknownthatcertainentriesaresignificantandtheremainingentriesareclosetoprobabilisticprocessesgenerateN×Cmatricesthatsatisfyzero.RIPwithhighprobability,thereisnopracticalalgorithmforTheworkofDonoho[9]andofCand`es,RombergandTaoverifyingwhetheragivensen

5、singmatrixΦhasthisproperty,[10],[2],[11]providesfundamentalinsightintothegeometrycrucialforthefeasibilityofthestandardrecoveryalgorithms.Incontrastthispaperprovidessimplecriteriathatguaranteethatofsensingmatrices.Thisgeometryisexpressedbye.g.theadeterministicsensingmatrixsatisfyin

6、gthesecriteriaactsasaRestrictedIsometryProperty(RIP),formulatedbyCand`esandnearisometryonanoverwhelmingmajorityofk-sparsesignals;Tao[10]:asensingmatrixsatisfiesthek-RestrictedIsometryinparticular,mostsuchsignalshaveauniquerepresentationinPropertyifitactsasanearisometryonallk-sparse

7、vectors;tothemeasurementdomain.Probabilitystillplaysacriticalrole,ensureuniqueandstablereconstructionofk-sparsevectors,itbutitentersthesignalmodelratherthantheconstructionofthesensingmatrix.AnessentialelementinourconstructionisissufficientthatΦsatisfy2k-RIP.WhenN/Cand/ork/Narethatw

8、erequirethecolumnsofthesensingmat

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。