论文部分内容阅读
本文建出一种应用近红外光谱技术鉴别野生台蘑的新方法。使用FieldSpec3便携式近红外光谱仪对包括野生台蘑在内的13种蘑菇进行漫反射光谱采集。将采集的数据经小波去噪后,对可见与近红外光谱(350～2500nm)进行峰谷筛选,所得峰谷集经主成分分析降维,取方差贡献率大于99.9%的5个主成分作为BP神经网络的输入值,建立数学模型。该模型在偏差±0.05内,对未知样本正确识别率为100%。本结果表明利用近红外漫反射光谱可以很好地鉴别野生台蘑。
In this paper, a new method for identification of wild Taiwanese by near-infrared spectroscopy was developed. Thirteen species of mushrooms including wild Tricholoma were collected by diffuse reflectance spectroscopy using a FieldSpec3 portable near-infrared spectrometer. After the data were denoised by wavelet, the peak and bottom of the visible and near infrared spectra (350 ~ 2500nm) were screened by principal component analysis, and the five principal components whose variance contribution rate was more than 99.9% BP neural network input values, the establishment of mathematical models. The model is within ± 0.05 deviation and the correct recognition rate for unknown samples is 100%. The results show that the use of near-infrared diffuse reflectance spectroscopy can be very good identification of wild Taiwan mushrooms.