基于近红外高光谱图像分析的麦粒硬度分类研究  

Hardness Classification of Wheat Kernel Based on Near-infrared Hyperspectral Imaging Technology

作　　者：张红涛[1] 田媛[1] 孙志勇[1] 母建茹 阮朋举 侯栋宸 

机构地区：[1]华北水利水电大学电力学院,河南郑州450011

出　　处：《河南农业科学》2015年第4期181-184,共4页Journal of Henan Agricultural Sciences

基　　金：国家自然科学基金项目(31101085);河南省基础与前沿技术研究计划项目(122300410145);河南省高等学校青年骨干教师资助计划项目(2011GGJS-094);华北水利水电大学教学名师培育项目(2014108);华北水利水电大学2014年大学生创新创业计划项目(HSCX2004143)

摘　　要：为对小麦硬度进行自动检测,采集不同硬度小麦品种的近红外高光谱图像,将光谱数据经过求导处理后,提取950~1 645 nm有效光谱区间数据,然后经过多元散射校正,建立偏最小二乘判别分析（PLS-DA）模型。采用120粒小麦对模型进行训练,剩余的90粒进行检验,总体上模型分类准确率为99.63%。表明,采用近红外高光谱成像技术对单籽粒小麦硬度进行分类是可行的。As the hardness of wheat may greatly influence the milling process,it is necessary to activate automatic detection of wheat hardness. To prepare for the research,the near-infrared hyperspectral images of wheat with different hardness were collected. The data were processed by derivation,and those in spec-tral range between 950—1 645 nm effective were extracted,after multiplicative scatter correction,with which a partial least squares discriminant analysis model（PLS-DA） was built. During the experiment,120 wheat kernels were used to train the model,and the remaining 90 kernels were used to predict. Conse-quently,the accuracy rate of the model was 99. 63% . The results showed that it was feasible to classify the hardness of wheat kernel based on near-infrared hyperspectral imaging technology.

关 键 词：近红外高光谱图像 光谱分析 偏最小二乘判别分析 小麦硬度 分类 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] S512.1[自动化与计算机技术—计算机科学与技术]