[关键词]
[摘要]
本文以牛奶中蜡样芽孢杆菌污染度的检测为研究对象,应用高光谱成像技术,结合图像处理技术、光谱分析技术和化学计量技术,探索了构建牛奶中蜡样芽孢杆菌污染度预测模型的可行性。应用图像处理技术选取样品分析区域,采用能量值(Energy)纹理特征函数降维处理高光谱数据并得到其特征值,建立了蜡样芽孢杆菌的PLS预测模型,模型中校正集与预测集的相关系数分别为0.9231和0.9054,RMSEC(校正均方根误差)和RESEP(预测均方根误差)分别为0.7336和0.8139。分析表明,PLS预测模型仅能对牛奶中蜡样芽孢杆菌进行高低浓度的鉴定。因此,提出了二维相关技术结合N-PLS构建预测模型的方法,N-PLS预测模型中校正集与预测集的相关系数分别为0.9999和0.9984,RMSEC和RESEP分别为0.022和0.0928。结果表明,N-PLS预测模型精度较高,能够对牛奶中蜡样芽孢杆菌实现定量分析。
[Key word]
[Abstract]
In this paper, a new quantitative detection method based on hyperspectral imaging technology was proposed and studied, which was applied to test Bacillus Cereus in liquid milk. The feasibility of the prediction models for detection the content of Bacillus Cereus in milk was explored by image processing technology, spectral analysis technology and chemical metrology technology, based on the detection of the contamination degree of Bacillus cereus in milk. The image processing technology was used to select the sample area, and the energy value (Energy) texture feature was applied to reduce the dimensions of hyperspectral data to obtain the characteristic value based on the texture feature analysis. A PLS model was built to predict the content of Bacillus Cereus in milk. The correlation coefficients between the calibration set and the prediction set in the PLS prediction model were 0.92 and 0.91, and RMSEC and RESEP were 0.73 and 0.81, respectively. The results showed that the PLS prediction model could only identified the high and low concentration of Bacillus Cereus in the milk. Therefore, the two-dimensional correlation technology combined with N-PLS method was proposed, and the N-PLS prediction model was built. The correlation coefficients between the calibration set and the prediction set in the N-PLS prediction model were 0.99 and 0.99 respectively, and RMSEC and RESEP were 0.02 and 0.09, respectively. The results showed that the N-PLS model had higher accuracy and was able to achieve quantitative analysis of Bacillus Cereus in milk.
[中图分类号]
[基金项目]
天津市科技计划项目(13JCYBJC25700);天津农学院高校教师教育改革创新引导发展项目(20170201)