高光谱技术结合GLCM的油桃品种判别研究

Study on Nectarine Variety Discrimination by Hyperspectral Technology Combined with GLCM

为实现对同产地4个油桃品种分类,本研究采用420~1000 nm波段和900~1700 nm波段高光谱技术对其进行处理,通过GLCM提取得到可见/近红外波段的图像纹理特征值差异较显著,并建立PLS、LS-SVM、ELM模型对4类油桃品种进行分类判别研究。研究结果表明:对于可见/近红外波段下的图像纹理特征值采用PLS模型建立的判别精度最高,其整体模型的判别正确率达到81.49%,可为检测设备研究提供理论支撑。

In order to classify 4 different kind of nectarine cultivars from the same producing area,420~1000 nm and 900~1700 nm hyperspectral techniques were used in this study.The difference of texture features between visible and near-infrared obtained by GLCM were significant,PLS,LS-SVM and ELM models were established to classify 4 nectarine varieties.The results show that the precision of PLS model is the highest,and the discrimination accuracy was 81.49%,which can provide theoretical support for the research of detection equipment.