欢迎来到天天文库
浏览记录
ID:53570046
大小:644.90 KB
页数:7页
时间:2020-04-18
《基于多光谱成像选取四季豆叶片的特征波段-论文.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、激兴与光电子学进展51.011101(2014)Laser&0ptoelectronicsProgress~2014~中国激光》杂志社基于多光谱成像选取四季豆叶片的特征波段曹鹏飞李宏宁罗艳琳林立波许海滨冯洁云南师范大学物理与电子信息学院,云南昆明650500摘要在400~720nm波段范围,基于液晶可调谐滤波器(LCTF)和CMOS相机组合的多光谱成像系统,以四季豆叶片为研究对象每隔5nm进行成像。根据图像亮度信息法和波段指数法的相关原理,首先分别计算得到各波段四季豆叶片的波段指数值和可识别度;然后对四季豆叶片的波段指数值和可识别度进行排序,综合图像的灰度离散、亮度信息丰富和波段的相关性小
2、等特点,得出545、630、645、720、650和570nm波段有较大的波段指数值和较好的识别度;最后根据最小欧氏距离法和光谱角度匹配法分别对四季豆叶片的特征波段的分类精度予以计算,两种方法的分类精度分别为100.00%和83.33%,得出选取的特征波段对四季豆叶片具有较好的分类精度。因此,545、630、645、720、650和570nm波段可作为四季豆叶片的特征波段。关键词成像系统;特征波段;多光谱成像;四季豆叶片;分类精度中图分类号O439文献标识码Adoi:10.3788/L0P51.011101SelectionofFeatureBandsforPhaseolusVUlgari
3、sLeavesBasedonMulti_-SpectralImagingCaoPengfeiLiHongningLuoYanlinLinLiboXuHaibinFengJieSchoolofPhysicsandElectronicInformation,YunnanNormalUniversityKunming,Yunnan650500,ChinaAbstractMulti——spectralimagesofPhaseolusVU19arisleavesatthewavelengthrangeof400-720nmwithanintervalof5nmarecapturedbyusinga
4、multi—spectralimagingsystemwhichmainlyconsistsofliquidcrystaltunablefilter(LCTF)andCMOScamera.Firstly,accordingtotheprinciplesofimagebrightnessandbandindex,thevalueofbandindexandidentifiabilityforPhaseolusVUlgarisleavesarecalculatedrespectivelyamongeveryband.Then,throughsortingthevalueofbandindexa
5、ndidentifiabilityforPhaseolusvulgarisleaves,itcanbeconcludedthatbands545,630,645,720,650and570nmhavepreferableidentificationwithconsideringthecharacteristicsofdiscretegraylevelsandrichbrightnessofimagesandlittlecorrelationcoefficientamongdifferentbands.Finally,theclassificationaccuracyforPhaseolus
6、VU19rarisleavesiscalculatedaccordingtotheprinciplesofminimumEuclideandistanceandminimumspectralanglematching.TheclassificationaccuracyofcharacteristicbandsforPhaseolus∞钆19rarisleavesis100.OO%and83.33%separatelythroughusingthesetwomethods.Wecandrawaconclusionthatthesebandshaveidealclassificationacc
7、uracy.Therefore,bands545,630,645,720,650and570nmcanbeusedasfeaturebandsforPhaseolusVUarisleaves.Keywordsimagingsystems;featureband;multi—spectralimaging;PhaseolusVUlgarisleaves;classificationaccuracy0CIScodesl10.
此文档下载收益归作者所有