基于神经网络的木材光谱分类方法研究

Research on the Spectral Classification Method of Wood Based on Neural Network

在木材的生产、流通、利用及加工过程中,需对木材进行分类识别。本次实验选取了绿檀、黑胡桃、非洲檀香木三种木材作为实验对象,获取了太赫兹时域光谱信号并提取其吸收系数谱,构建了三层反向传播(Back Propagation,BP)神经网络分类模型,并进行了模型训练和测试。实验结果表明,借助BP神经网络,三种木材的分类准确率、精确率、召回率、F1得分等性能指标均取得了较好值。

In the production,distribution,utilization and processing of wood,wood needs to be classified and identified.In this experiment,three kinds of wood,namely green sandalwood,black walnut and African sandalwood,were selected as experimental objects,and the terahertz time-domain spectral signals were obtained and their absorption coefficient spectra were extracted.A three-layer BP neural network classification model was built,and model training and testing were carried out.The experiments show that with the help of back propagation(BP)neural network,the performance indicators such as classification accuracy,precision,recall,F1 score,etc.of the three kinds of wood have achieved good values.