针对茶叶品质无损检测时内外品质难以同时兼顾的问题,利用高光谱图像技术检测茶叶质量。设计一套基于光谱仪的高光谱图像系统采集数据;通过主成分分析,从海量数据中优选出三个波长段的特征图像;从每个特征图像中分别提取平均灰度级、标准方差、平滑度、三阶矩、一致性和熵等6个基于统计矩的纹理特征参量,每个样本共有18个特征变量;再通过主成分分析对这18个特征变量进行压缩,提取8个主成分因子建立基于反向传播神经网络的茶叶等级判别模型。模型训练时的总体回判识别率为97%;预测时总体识别率为94%。结果表明,高光谱图像技术可以用于茶叶质量等级水平的评判。

The tea quality level was estimated with hyperspectral imaging technology. A hyperspectral imaging system based on spectrometer was developed to perform acquisition of hyperspectral imaging data. The principal component analysis (PCA) was performed to select three optimal bands images. Then, six texture features (i.e., mean, standard deviation, smoothness, third moment, uniformity, and entropy) based on the statistical moment were extracted from each optimal band image, thus 18 variables for each tea sample in all. Finally, PCA was performed again to compress 18 features variables, and 8 principal components (PCs) were extracted as the input of back propagation (BP) neural net. Experimental results showed that discriminating rate was 97% in the training set and 94% in the prediction set. Overall results sufficiently demonstrate that the hyperspectral imaging technology can be used to estimate tea quality level.