基于图像特征的柑橘黄龙病视觉鉴别

摘要：近年来，柑橘黄龙病（（ Huanglongbing，HLB）严重威胁着柑橘产业的健康发展，及时准确地识别柑橘黄龙病具有 重要意义。高光谱具有光谱分析和图像处理的特点，在植物病害检测中表现出了巨大的优势。本文结合高光谱成像技术， 探索基于图像特征的柑橘黄龙病视觉鉴别方法。以轻度黄龙病、中度黄龙病、重度黄龙病、营养缺陷病和正常柑橘叶片 为研究对象，采用聚合酶链式反应（（ PCR）试验进行视觉区划验证。采用连续投影算法（（ SPA）对光谱提取变量，对图像 进行主成分分析（（ PCA）。选择敏感波长，分析特征图像，提取基于灰度共生矩阵（（ GLCM）的图像纹理特征，建立偏最 小二乘判别分析（PLS-DA）模型。讨论了不同纹理个数对模型的影响，结果表明当4个独立的纹理特征共计36个变量 作为输入时，模型分类效果最好，误判率较大，为3.12%，预测相关系数（（RP）较大，为0.98，预测均方根误差（（RMSEP） 较小，为0.32，主成分因子（PC）个数为16个。结果表明，基于GLCM的图像纹理特征结合PLS-DA模型可以实现柑 橘黄龙病的识别，为黄龙病的视觉判别研究提供重要的参考价值。

关键词：柑橘黄龙病；视觉判别；GLCM；图像纹理特征；PLS-DA

期刊：Accepted Manuscript

DOI：https://DOI.org/10.1016/j.vibspec.2019.04.001

设备：GaiaSorter

Visual Discrimination Of Citrus HLB Based On Image Features

Abstract：Citrus greening (Huanglongbing, HLB) has seriously threatened the healthy development of citrus industry in recent years, and it was great significance to identify citrus HLB timely and accurately. Hyperspectral has the characteristics of spectral analysis and image processing, which has shown great advantages in plant disease detection. This paper combined with hyperspectral imaging technology to explored the visual identification method of citrus HLB based on image features. Slight HLB, moderate HLB, serious HLB, nutritional deficiency and normal citrus leaves as research object, polymerase chain reaction (PCR) test was used to verify visual division. Variables were extracted by successive projections algorithm (SPA) for spectra and principal component analysis (PCA) for image. Selection of sensitive wavelengths, analysis of feature images and extraction for texture features of images based on grayscale co-occurrence matrix (GLCM) was used to conduct a partial least squares discriminant analysis (PLS-DA) model. The influence of different texture numbers on the model was discussed, which indicated that the model classification effect was the best when the 4 independent texture features were 36 variables as input, which resulted in greater misjudgment rate of 3.12%, higher correlation coefficient for prediction (RP) of 0.98, lower root mean square error for prediction (RMSEP) of 0.32, and number of principal component factors (PCs) of 16. The results highlighted that the image texture feature based on GLCM combined with PLS-DA model could realize the identification of citrus HLB, and provide important reference value for the visual discrimination research of HLB. Keyword：Citrus greening; Visual discrimination; GLCM; Image texture feature; PLS-DA