欢迎来到天天文库
浏览记录
ID:1136525
大小:5.77 MB
页数:8页
时间:2017-11-07
《一种基于单形体正化的高光谱数据全约束线性解混方法》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、第35卷第5期红外与毫米波学报Vol.35ꎬNo.52016年10月J.InfraredMillim.WavesOctoberꎬ2016文章编号:1001-9014(2016)05-0592-08DOI:10.11972/j.issn.1001-9014.2016.05.014一种基于单形体正化的高光谱数据全约束线性解混方法1ꎬ2ꎬ3∗1ꎬ21ꎬ24许宁ꎬ耿修瑞ꎬ尤红建ꎬ曹银贵(1.中国科学院空间信息处理与应用系统技术重点实验室ꎬ北京100190ꎻ2.中国科学院电子学研究所ꎬ北京100190ꎻ3.中国科学院大学ꎬ北京100049ꎻ4.中国地质大学ꎬ北京100083)摘要:在端元已
2、知情况下ꎬ线性混合模型的非负约束最小二乘无闭式解ꎬ需要多次迭代得收敛最优解ꎬ时间复杂度高.通过高光谱数据凸面几何特性分析ꎬ指出当数据为正单形体时ꎬ可经有限步骤快速得线性混合模型最优解.据此提出一种单形体正化的高光谱数据全约束线性解混方法ꎬ据已知端元进行单形体正化ꎬ采用和为一约束求解丰度系数ꎬ最后迭代剔除丰度负值端元得全约束解.实验结果表明该方法可获得传统全约束解一致的丰度估计ꎬ且效率大大提升.关键词:高光谱数据ꎻ光谱解混ꎻ端元白化ꎻ单形体正化ꎻ全约束最小二乘中图分类号:TP394.1文献标识码:AAfullyconstrainedlinearunmixingmethod:Simp
3、lexregularizationforhyperspectralimagery1ꎬ2ꎬ3∗1ꎬ21ꎬ24XUNingꎬGENGXiu ̄RuiꎬYOUHong ̄JianꎬCAOYin ̄Gui(1.KeyLaboratoryofTechnologyinGeo ̄spatialInformationProcessingandApplicationSystemꎬIECASꎬBeijing100190ꎬChinaꎻ2.InstituteofElectronicsꎬChineseAcademyofSciencesꎬBeijing100190ꎬChinaꎻ3.UniversityofChine
4、seAcademyofSciencesꎬBeijing100049ꎬChinaꎻ4.SchoolofLandScienceandTechnologyꎬChinaUniversityofGeosciencesꎬBeijing100083ꎬChina)Abstract:Withaprioriinformationoftheknownendmembersinhyperspectralimageꎬthereisnoclosed ̄formsolutionofLeastSquare(LS)methodforlinearmixingmodelundertheAbundanceNon ̄negat
5、ivityConstraint(ANC).SomanyiterationswhichmayresultinbigcomputationalcomplexityareneededinthetraditionalFullyConstrainedLS(FCLS)methodstoobtaintheoptimalsolution.Inthispaperꎬananalysisofimpactsonabundanceestimationofhyperspectalimageindifferentsimplexshapeswasimplementedandafullyconstrainedli
6、nearunmixingmethodbasedonsimplexregulariza ̄tionwasproposedwhichcouldgetoptimalsolutionunderlimitediterationwhenthehyperspectralimagewasspannedintoaregularsimplex.Theproposedmethodwascarriedoutbythreesteps.Firstlyꎬthesimplexofhyperspectralimagewasregularizedbytheknownendmembers’whiteningmatrix
7、.Second ̄lyꎬtheanalyticalsolutionofabundancecoefficientswasobtainedunderAbundanceSum ̄to ̄oneCon ̄straint(ASC).ThenforeverypixelꎬtheFCLSsolutionwasachievedbyeliminatingtheendmemberswithnegativeabundancecoefficientsandsolvingtheASCequationiterativ
此文档下载收益归作者所有