欢迎来到天天文库
浏览记录
ID:55808361
大小:962.98 KB
页数:6页
时间:2020-06-03
《高分辨光学压缩光谱成像方法与实验研究.pdf》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、第34卷第1期光学学报Vol_34,No.12014年1月ACTAOPTICASINICAJanuary,2014高分辨光学压缩光谱成像方法与实验研究陈宇,匣’3’4周建康’4陈新华季轶群沈为民。’4苏州大学现代光学技术研究所,江苏苏州215006。江苏省先进光学制造技术重点实验室,江苏苏州215006。江苏省现代光学技术重点实验室,江苏苏州215006教育部现代光学技术重点实验室,江苏苏州215006摘要光学压缩光谱成像是融合了压缩感知原理的新型光谱成像技术,具有降低数据采集量、能对景物实施凝视拍摄、提高信噪比等优点。考虑到采样质量对最终成像质量的影响
2、,在现有成像系统中均采用采样间隔与调制间隔匹配的方法,但此方法降低了系统的采样率,损耗了原始光谱分辨率。针对上述成像方法缺陷,克服采样间隔和调制间隔匹配的成像系统设计要求,所设计实验装置使其光谱分辨率理论值提高至原先方法的3倍以上,并对最优化方法进行改进,在正则化函数中增加表征数据光谱维连续性的变差项,增强数据重建可控性及可靠性。实验结果表明,新方法下实验装置的光谱维通道数提升,各波段图像和特定位置光谱曲线能精确反映目标物的空间特性和光谱特性。关键词光谱成像;压缩感知;最优化方法;采样率;光谱分辨率中图分类号0433.4文献标识码Adoi:10.3788
3、/AOS2o1434.0¨l005ResearchonPrincipleandExperimentationofHigh-ResolutionOpticalCompressiveSpectralImagingChenYuheng,。'。,ZhouJiankang,’。,ChenXinhua,,。,JiYiqun,,。,ShenWeimin’。'。,/nstituteofModernOpticalTechnologies,SoochowUniversity,Suzhou,Jiangsu215006,China。KeyLaboratoryofAdvanced
4、OpticalManufacturingTechnologiesofJiangsuProvince,Suzhou,Jiangsu215006,China。KeyLaboratoryofModernOpticalTechnologiesofJiangsuProvince,Suzhou,Jiangsu215006,ChinaKeyLaboratoryofModernOpticalTechnologiesofEducationMinistryofChina,Suzhou,Jiangsu215006,ChinaAbstractOpticalcompressive
5、spectralimagingmethodisanovelspectralimagingtechniquethatdrawsintheinspirationofcompressedsensing,whichhasthefeaturessuchasreducingacquisitiondataamount,realizingsnapshotimagingforcertainscenery,increasingsignaltonoiseratioandSOon.Consideringtheinfluenceofthesamplingqualityontheu
6、ltimateimagingquality,matchingthesamplingintervalwiththemodulationintervalintheformerreportedimagingsystem,whilethedepressedsamplingrateleadstothelossontheoriginalspectralresolution.Toovercomethattechnicaldefect,thedemandforthematchingbetweensamplingintervalandmodulationintervali
7、Sdisposedandthespectralresolutionofthedesignedexperimenta1deviceincreasesmorethanthreefoldcomparingwiththatofthepreviousmethod.Optimizationmethodisimprovedandavariationtermthatrepresentsthespectral—dimensioncontinuousnessofthedataiSaddedtotheregularizationfunction,whichenhancesth
8、econtrollabilityandreliabilityforthedata
此文档下载收益归作者所有