摘要
针对马铃薯黑心病不易检测,提出马铃薯黑心病的光学无损检测方法,并比较了马铃薯黑心病的漫反射光谱和透射光谱检测方法。通过高光谱图像采集系统、透射光谱采集系统和傅里叶变换近红外光谱仪获取合格马铃薯与黑心病马铃薯的可见/近红外漫反射光谱、可见/近红外透射光谱以及近红外漫反射光谱,并采用偏最小二乘-线性判别分析方法建立马铃薯黑心病的识别模型。透射光谱采集系统采集的可见/近红外透射光谱所建模型的判别正确率最高,对测试集样本的识别正确率为98.46%;高光谱图像采集系统获取的可见/近红外漫反射光谱经二阶导与标准化组合预处理后所建模型对测试集样本的识别正确率为92.31%;傅里叶变换近红外光谱仪获取的漫反射光谱经标准正态变量变换与标准化组合预处理后所建模型对测试集样本的识别正确率90.77%。试验结果表明:采用光谱检测马铃薯黑心病,透射光谱系统优于高光谱成像系统,高光谱成像系统优于傅里叶近红外光谱仪。研究结果为马铃薯内部缺陷的光谱定性判别及便携式仪器的研制提供了参考。
The objective of this research was to compare transmission and diffuse reflection modes of visible （VIS）/near infrared （NIR） spectroscopy for detecting black heart in potatoes. Vis/NIR spectra, Vis/NIT spectra and NIR spectra were acquired using hyperspectral image acquisition system, portable transmission spectrum acquisition system and FT-NIR spectrometer, respectively. Partial Least Squares-Linear Discriminant Analysis （PLS-LDA） method was employed to classify the potatoes with or without black heart. Better results were obtained based on Vis/NIT spectra than those on diffuse reflection spectra. The classifying correct rate was 98.46% by using Vis/NIT spectra. The classifying correct rates were 92.31% and 90.77% respectively acquired by hyperspectral image spectrum acquisition system and FT-NIR spectrometer. The results indicated that Vis/NIT spectra method was feasible for detecting black heart in potatoes and transmission mode was better than reflection mode for internal disorder detection. The research can provide references for potato internal defect detection and portable instrument development based on spectroscopy technique.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2012年第11期237-242,共6页
Transactions of the Chinese Society of Agricultural Engineering
基金
湖北省自然科学基金重点项目(2011CDA033)
高等学校博士学科点专项科研基金(20090146110018)
关键词
近红外光谱
模型
优化
马铃薯
黑心病
高光谱
漫反射光谱
透射光谱
near infrared spectroscopy, models, optimization, potato, black heart, hyperspectral imaging, diffuse reflection spectrum, transmission spectrum
