资源描述：
《Phase unwrapping for large SAR interferograms statistical segmentation and generalized network models.pdf》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、IEEETRANSACTIONSONGEOSCIENCEANDREMOTESENSING,VOL.40,NO.8,AUGUST20021709PhaseUnwrappingforLargeSARInterferograms:StatisticalSegmentationandGeneralizedNetworkModelsCurtisW.ChenandHowardA.Zebker,Fellow,IEEEAbstract—Two-dimensional(2-D)phaseunwrappingisaweproposea
2、techniqueforapplyingChenandZebker’s[5]keystepintheanalysisofinterferometricsyntheticaperturestatistical-cost,network-flowphase-unwrappingalgorithmradar(InSAR)data.Whilechallengingeveninthebestof(SNAPHU)tolargedatasets.Weusetheterm“large”tocircumstances,thispro
3、blemposesuniquedifficultieswhenthedescribeinterferogramsapproachingorexceedingthelimitsofdimensionsoftheinterferometricinputdataexceedthelimitsofone’scomputationalcapabilities.Inordertodealwithsuchtypicalcomputerresourcesavailableforunwrappingthem.cases,weprop
4、oseatechniqueforapplyingthestatistical-cost,net-Anobviousstrategyforcircumventingtheseresourcelimi-work-flowphase-unwrappingalgorithm(SNAPHU)ofChenandtationsistoadoptanapproachinwhichalargeinterferogramZebkertolargedatasets.Specifically,weintroduceamethodology
5、istreatedasamosaicofsmallerinterferogramsor“tiles.”Thewherebyalargeinterferogramispartitionedintoasetofseveraltilescanbeunwrappedinsequencetoreducetherequiredsmallertilesthatareunwrappedindividuallyandthendividedfurtherintoindependent,irregularlyshapedreliable
6、regions.memory,ortheycanbeunwrappedinparalleltoreducetheTheseregionsaresubsequentlyassembledintoafullunwrappedrequiredexecutiontime[4],[6].Unfortunately,becauseofsolution,withthephaseoffsetsbetweenregionscomputedinaboundaryeffects(seebelow),smallerinterferogra
7、msareoftensecondaryoptimizationproblemwhoseobjectiveistomaximizemoredifficulttounwrapcorrectlythanlargerinterferograms.theaposterioriprobabilityofthefinalsolution.AsthissecondaryMoreover,thetaskofcombiningindividuallyunwrappedtilesproblemassumesthesamestatisti
8、calmodelsasemployedintheinitialtile-unwrappingstage,thetechniqueresultsinasolutionisnotstraightforward.Consequently,amosaicoftilesolutionsthatapproximatesthesolutionthatwouldhavebe
