【摘要】 为实现杏的品种判别，采集“6-1”杏、网红杏、晋梅杏和扁杏4个品种的光谱数据，选用SG、MA、MF、Baseline、SNV、MSC 6种预处理方法，并建立偏最小二乘模型（PLSR）对杏的品种进行判别。结果表明，MF预处理所建模型效果最佳，预测集的决定系数为0.840 2，均方根误差为0.446 7。为简化模型，对MF预处理后的光谱数据分别采用回归系数法（RC）和连续投影算法（SPA）选取特征波长建模，最优模型SPA-PLSR的测试集的总判别率为84.44%。该研究可为杏的品种判别提供理论参考。更多还原

【Abstract】 In order to realize the variety discrimination of apricot,the spectral data of“6-1”,Wanghong apricot,Jinmei apricot and flat apricot were collected,and six preprocessing methods of SG,MA,MF,Baseline,SNV and MSC were selected,and a partial least squares model(PLSR) was established to distinguish the apricot varieties. The results showed that the model built by MF preprocessing performs the best,with the prediction set having a coefficient of determination of 0.840 2 and the RMSE of 0.446 7. To simplify the model,regression coefficient method(RC) and continuous projection algorithm(SPA) were used for the spectral data. The total accuracy of the optimal model SPA-PLSR was 84.44%. This study could provide a theoretical reference for the variety discrimination of apricot.更多还原