基于PLS-DA烟叶成熟度判别模型研究

Research on Discriminate Model of Tobacco Leaves Maturity Based on PLS-DA

目前烟农根据烟叶外观特征判别烟叶成熟度,具有较强的主观性强,影响烟草烘烤工艺的制定及烤后烟叶的品质。该研究以贵州省云烟87中部烟叶为研究对象,首先利用搭建的高光谱信息采集系统获取了烟叶的高光谱图像;然后应用软件SpecView提取烟叶样本感兴趣区域ROI的平均光谱,采用Savitzky-Golay平滑法预处理降噪后,将308个样本随机分208个作为训练集和100个作为测试集;最后计算了6个光谱指数及4个可见光光谱参数平均值,选择其中具有线性变化规律的7个(SIPI、mNDVI705、VOG3、I1、I2、I3、I4)光谱参数,创建了偏最小二乘法判别分析(PLS-DA)模型,模型对烟叶成熟度判别准确率达到98%。研究结果表明,将高光谱图像技术结合PLS-DA能够准确地判别烟叶的成熟度。

At present,the strong subjectivity of tobacco growers to distinguish the maturity grades of tobacco leaves according to the their fresh appearance characteristics may affect the formulation of tobacco baking technology and the quality of baked tobacco leaves. In this study,the fresh tobacco leaves in the middle of Yunyan 87 in Xifeng County(Central Guizhou),Anlong county(southwest Guizhou),Daozhen county(North Guizhou)and Weining County(Northwest Guizhou)are taken as the research objects. Above all,the hyperspectral images of fresh tobacco are obtained by the hyperspectral information acquisition system,and the Monochrome correction of the hyperspectral image is carried out. Then,the average spectrum in the region of interest(ROI)of tobacco samples is extracted by the Specview and denoise pretreatment is applied by the Savitzky-Golay smoothing method to randomly divide 308 samples into 208 modeling sets and 100 prediction sets. Finally,six spectral indexes and four average values of visible spectral parameters are calculated. Seven spectral parameters(SIPI、mNDVI705、VOG3、I1、I2、I3、I4)with linear variation regulation are selected to form multispectral parameter vector group as input parameters,and a partial least squares discriminant analysis(PLS-DA)model is established,which is fulfilled with 98% accuracy for maturity grades discrimination of fresh tobacco leaves.The results show that the combination of hyperspectral image technology and PLS-DA can be utilized to distinguish the maturity grades of fresh tobacco leaves.