基于太赫兹时域光谱技术的红木检测方法

发布时间：2018-05-11 02:13

本文选题：光谱学 + 太赫兹时域光谱　；参考：《光学学报》2017年02期

【摘要】：提出了一种基于随机森林预测模型的太赫兹时域光谱的木材鉴别方法。对4种木材(2种红木、2种非红木)在0.2~1.2THz频率范围的吸收光谱的差异进行分析;对得到的光谱吸光度数据进行主成分分析的数据降维处理,并提取方差贡献率最高的五种主成分(总贡献率高达99.65%);将其代入随机森林预测模型预测鉴别红木的真伪,得出相应训练集和测试集的识别率。实验结果表明,与传统的支持向量机预测模型和单一决策树模型比较,使用时域光谱技术结合随机森林预测模型能够得到更高的识别率,识别率可达91.25%,能够准确对红木和非红木进行检测。
[Abstract]:A method of wood identification based on THz time domain spectrum based on stochastic forest prediction model is proposed. The differences of absorption spectra in the frequency range of 0.2~1.2THz were analyzed, and the data of principal component analysis (PCA) were reduced. The absorption spectra of 4 species of wood, 2 species of red wood and 2 species of non-mahogany were analyzed in the frequency range of 0.2~1.2THz, and the spectral absorbance data were reduced by principal component analysis (PCA). The five principal components with the highest contribution rate of variance (the total contribution rate is as high as 99.65g) are extracted and replaced into the stochastic forest prediction model to predict and identify the true or false of mahogany, and the recognition rates of the corresponding training set and test set are obtained. The experimental results show that compared with the traditional support vector machine prediction model and the single decision tree model, the time domain spectroscopy combined with the stochastic forest prediction model can achieve a higher recognition rate. The recognition rate can reach 91.25, and can accurately detect redwood and non-mahogany.
【作者单位】：桂林电子科技大学电子工程与自动化学院;广西高校光电信息处理重点实验室;
【基金】：国家自然科学基金(61565004) 广西自然科学基金(2013GXNSFDA019002,2014GXNSFGA118003) 桂林市科学研究与技术开发课题(20140127-1,20150133-3)
【分类号】：O433.4

【相似文献】