基于近红外高光谱图像的黄瓜叶片色素含量快速检测  

Rapid Detection of Cucumber Leaves Pigments Based on Near Infrared Hyper-spectral Image Technology

作　　者：邹小波[1] 陈正伟[1] 石吉勇[1] 黄晓玮[1] 张德涛[1] 

机构地区：[1]江苏大学食品与生物工程学院,镇江212013

出　　处：《农业机械学报》2012年第5期152-156,共5页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金资助项目(60901079);国家高技术研究发展计划(863计划)资助项目(2008AA10Z208)

摘　　要：利用高光谱图像技术和高效液相色谱法(HPLC)快速检测了新鲜黄瓜叶中叶绿素a、叶绿素b、β-胡萝卜素和叶黄素4种色素含量。采集了120片黄瓜叶的近红外高光谱图像数据以及用HPLC精确测定黄瓜叶中色素含量;提取高光谱图像中50×50像素感兴趣区域(ROI)的平均光谱与4种色素含量分别建立偏最小二乘(PLS)预测模型;为了提高模型的稳定性和预测精度,分别采用区间偏最小二乘(iPLS)、向后区间偏最小二乘(BiPLS)和联合区间偏最小二乘(SiPLS)对各种色素对应的特征波段进行优选,同时对光谱划分数进行了优化。结果表明BiPLS和SiPLS对应模型的预测效果较好,对叶绿素a、叶绿素b、β-胡萝卜素和叶黄素4种色素的预测集相关系数RP分别为0.825 7、0.813 4、0.811 6、0.826 2。Rapid detection the content of chlorophyll a,chlorophyll b,β-carotene and lutein on fresh cucumber leaves by using hyper-spectral image and high performance liquid chromatography（HPLC） was done.One hundred and twenty hyper-spectral images of cucumber leaves were collected by near infrared hyper-spectral camera,then using HPLC to accurate detect the content of chlorophyll a,chlorophyll b,β-carotene and lutein.After hyper-spectral images have been corrected,a 50×50 pixels region as region of interest（ROI） was defined.Average spectrum from ROI was extracted and prediction model of partial least squares（PLS） with the content of four pigments was built.In order to advance the stability and prediction accuracy,interval partial least squares（iPLS）,backwards interval partial least squares（BiPLS） and synergy interval partial least squares（SiPLS） algorithm was used to select diagnostic bands of each pigments after optimized the number of spectrum interval.The result showed that using BiPLS and SiPLS would get the best PLS model,the best prediction coefficient of chlorophyll a,chlorophyll b,β-carotene and lutein is 0.825 7,0.813 4,0.811 6 and 0.826 2,respectively.

关 键 词：黄瓜叶片 色素 高光谱图像 高效液相色谱 偏最小二乘法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] Q94-3[自动化与计算机技术—计算机科学与技术]