基于高光谱成像技术的茄子叶片灰霉病早期检测  

Early detection of gray mold on eggplant leaves using hyperspectral imaging technique

作　　者：冯雷[1] 张德荣[2] 陈双双[1] 冯斌[3] 谢传奇[1] 陈佑源[4] 何勇[1] 

机构地区：[1]浙江大学生物系统工程与食品科学学院,浙江杭州310058 [2]浙江大学宁波理工学院,浙江宁波315000 [3]全国农业展览馆,北京100026 [4]浙江大学生物技术研究所,浙江杭州310058

出　　处：《浙江大学学报（农业与生命科学版）》2012年第3期311-317,共7页Journal of Zhejiang University:Agriculture and Life Sciences

基　　金：国家高技术研究发展计划资助项目(2011AA100705);国家自然科学基金资助项目(61075017);浙江省自然科学基金资助项目(Y5090044);浙江省重大科技专项重点农业资助项目(2009C12002)

摘　　要：为建立基于高光谱成像技术的茄子叶片灰霉病早期检测方法,利用高光谱成像系统获取120个茄子叶片在380～1 031nm范围的高光谱图像数据,通过主成分分析(PCA)对高光谱数据进行降维,并从中优选出3个特征波段下的特征图像,截取200×150的感兴趣区域图像(ROI),并从每幅特征图像中分别提取均值、方差、同质性、对比度、差异性、熵、二阶矩和相关性等8个基于灰度共生矩阵的纹理特征变量,通过连续投影算法(SPA)提取13个特征变量,利用最小二乘支持向量机(LS-SVM)构建茄子叶片灰霉病早期鉴别模型,模型判别准确率为97.5%。说明高光谱成像技术可以用于茄子叶片灰霉病的早期检测。Early detection of gray mold on eggplant leaves using hyperspectral imaging technique was proposed. Hyperspectral images of 120 eggplant samples were captured by hyperspectral imaging system, and the spectral region was from 380 to 1 031 nm. The pictures on three feature wavelengths were selected by principal component analysis （PCA）, which was a good method to reduce the dimension of hyperspectral data. Eight feature variables were extracted by texture analysis based on gray level co-occurrence matrix （GLCM） after choosing the region of interest （ROI） of 200 ）〈 150, which were mean, variance, homogeneity, contrast, dissimilarity, entropy, second moment, correlation respectively, thus 24 feature variables in total for three feature images. Successive projections algorithm （SPA） was executed on 24 feature variables, 13 feature variables in which were extracted as the input of the least square support vector machines （LS-SVM） model, and the accurate rate of the model was 97.5 %. It is showed that it is feasible for early detection of gray mold on eggplant leaves by hyperspectral imaging technique.

关 键 词：高光谱成像技术 灰霉病 最小二乘支持向量机 连续投影算法 主成分分析 茄子 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] S436.411[自动化与计算机技术—计算机科学与技术]