欢迎来到天天文库
浏览记录
ID:15754371
大小:39.00 KB
页数:16页
时间:2018-08-05
《基于高光谱成像技术的鸡肉菌落总数快速无损检测》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、基于高光谱成像技术的鸡肉菌落总数快速无损检测关键字:基于高光谱成像技术的鸡肉菌落总数快速无损检测本文为Word文档,感谢你的关注! 摘要:以市售新鲜冷藏(4℃)鸡胸肉为研究对象,采集鸡胸肉表面的高光谱(400~1100nm)图像信息,采用偏最小二乘回归(partialleastsquareregression,PLSR)建立菌落总数预测模型,采用不同预处理方法提高模型的预测准确性和稳健性,实现快速无损检测生鲜鸡胸菌落总数的目的。结果表明:标准变量变换(standardnormalizedvariate,SNV)预处理后,模型性能最佳。模型的校正标准差(sta
2、ndarderrorofcalibration,sEC)和验证标准差(standarderrorofprediction,sEP)分别为0.40和0.57,sEP/sEC为1.08,校正集相关系数(correlationcoefficientofprediction,RC)和验证集相关系数(correlationcoefficientofprediction,RP)分别为0.93和0.86;且应用最佳模型可有效预测样品菌落总数的分布地图。 关键词:鸡肉;菌落总数;高光谱成像;图像预处理;偏最小二乘法(PLSR);无损检测 RapidNon-Destruct
3、iveDetectionofTotalBacterialCountinChickenUsingHyperspectralImaging LIWencai1,LIUFei1,TIANHanyou1,ZOUHao1,WANGHui1,ZHANGZhenqi1,ZHENGXiaochun2,LIYongyu2, LIJiapeng1,QIAOXiaoling1,* (1.BeijingKeyLaboratoryofMeatProcessingTechnology,ChinaMeatResearchCenter,Beijing100068,China; 2.C
4、ollegeofEngineering,ChinaAgriculturalUniversity,Beijing100083,China) Abstract:Inordertodeveloparapidandnon-destructivemethodtopredicttotalbacterialcountinchickenbreastsbyusinghyperspectralimagingtechnology,83chickenbreastsamplesrefrigeratedat4℃werecollectedfromlocalsupermarketand63
5、ofthesewereusedascalibrationsamples.Thehyperspectralscatteringimageofeachsamplewascollectedbyusinghyperspectralimagingsysteminthewavelengthrangeof400?1100nm.Variousalgorithmcombinationswereusedtopreprocessthehyperspectralinformationofthesamplestoenhancetheperformanceofthemodeldevelo
6、pedbyusingpartialleastsquareregression(PLSR)algorithm.Basedonthepredictiveaccuracyandstabilityofthemodel,theefficiencyofdifferentalgorithmcombinationsforspectralpreprocessingwereevaluatedanddiscussed.Theresultsshowedthattheoptimalmodelperformancewasachievedbypreprocessingofthehypers
7、pectralinformationwithstandardnormalizedvariate.Thestandarderrorofcalibration(sEC)andstandarderrorofprediction(SEP)ofthemodelwere0.40and0.57,respectively.Thecorrelationcoefficientsofcalibration(RC)andprediction(RP)were0.93and0.86,respectively.Theoptimalmodelallowedeffectivepredictio
8、nofdistributionmaps
此文档下载收益归作者所有