论文部分内容阅读
以喷洒不同浓度杜邦万灵的香梨作为研究对象,探讨了应用高光谱成像技术检测香梨表面农药残留的方法。运用376~1051nm高光谱成像系统采集200个香梨的高光谱图像,其中120个香梨为建模集,80个香梨为预测集。运用多元散射校正(MSC)对光谱数据进行预处理,然后采用连续投影算法(SPA)提取了11个特征波长。基于处理后的光谱数据,分别运用多元线性回归法(MLR)和主成分回归法(PCR)两种算法分别建立农药残留检测的模型,比较两种模型的结果。通过比较,采用MLR建立的农药残留检测模型效果较优,其校正集相关系数(Rc)为0.973,校正均方根误差(RMSEC)为0.260,预测的正确率可以达到91.7%,对较低浓度残留的预测正确率达到80%。研究表明,应用高光谱成像技术可以成功地检测香梨表面农药残留,并且对低浓度检测也有很好的效果。
In order to study the pesticide residues in pear surface by using hyperspectral imaging technology, we sprayed pear with different concentrations of pompom. Hyperspectral images of 200 pear cultivars were collected using a 376 ~ 1051 nm hyperspectral imaging system. Among them, 120 pear cultivars were model sets and 80 pear cultivars were predicted sets. The spectral data were preprocessed by using multivariate scatter correction (MSC), and 11 characteristic wavelengths were extracted by continuous projection algorithm (SPA). Based on the processed spectral data, a multi-linear regression (MLR) and principal component regression (PCR) algorithm were used to establish the pesticide residue detection model respectively, and the results of the two models were compared. The results showed that the correlation coefficient (Rc) of calibration set was 0.973, the root mean square error of correction (RMSEC) was 0.260 and the prediction accuracy was 91.7% Residual prediction accuracy rate of 80%. Studies have shown that the application of hyperspectral imaging technology can successfully detect pear pesticide residues on the surface, and low concentrations of detection also have good results.