考虑直径影响的苹果霉心病透射光谱修正及检测

Detection Method of Moldy Core in Apples Using Modified Transmission Spectrum Based on Size of Fruit

摘要:

针对苹果霉心病近红外透射光谱信息受果实直径影响的难题，提出了一种能够修正果实直径对透射光谱影响的方法。基于透射光谱采集平台获取327个红富士苹果的可见/近红外光谱（350～1100nm）信息，采用电子游标卡尺获取其直径（光程）信息。以直径为80mm健康苹果的平均光谱作为参考光谱，将327个苹果的光谱与参考光谱进行比较，结合直径信息利用公式求得透射光在果实内的衰减系数P，用衰减系数P进行透射光谱的修正。修正后光谱建立支持向量机（SVM）模型、误差反向传播神经网络（BP-ANN）模型，并与修正前原始光谱建立模型进行对比。实验结果表明，应用此光谱修正方法能够显著提高模型判别精度，其中应用SVM算法对修正后的光谱建立模型效果最好，对训练集和测试集的判别准确率分别为99.34%和90.20%，相对于原始光谱建立的模型判别准确率分别提高了7.84和5.89个百分点。基于此方法修正果实直径对于透射光谱的影响是可行的，构建的模型能够实现苹果霉心病的准确判别。

Abstract:

Currently, the near infrared transmission spectrum of moldy core in apples was seriously affected by the size of fruit. In order to solve the problem, a transmission spectrum correction method based on size of fruit was proposed. A spectrum acquisition platform was constructed to acquire the transmission spectra (350~1100nm) of 327 Fuji apples and their diameters were measured with a vernier caliper. The spectrum of healthy apples with diameter of 80mm was used as reference. Comparing the spectrum of 327 apples with the reference spectrum, a formula was built. The attenuation index of transmitted light in the fruit can be easily found by using the formula and diameters. Then the transmission spectrum was modified with the help of attenuation index. Error back propagation artificial neural networks (BP-ANN) and support vector machine (SVM) measurement model were established based on corrected spectrum and original spectrum. The results showed that the accuracy of the models based on corrected spectrum was much higher than those of the others, and its recognition accuracy rate reached 99.34% for the training set and 90.20% for the test set. The recognition rate of the model was 7.84 and 5.89 percentage points higher than that of the original spectrum. The results showed that the effect of the size on transmission spectra can be corrected by this method, and the method had high identification accuracy. Meanwhile, the results would provide theoretical basis for the development of online detection of internal quality in apples and provide a new idea for the study of internal disease detection models for different agricultural products.