生菜叶片镉含量高光谱预测模型  

Hyperspectral Prediction Model of Cadmium Content in Lettuce Leaves

作　　者：李君妍 张跃春 LI Junyan, ZHANG Yuechun(School of Electrical and Information Engineering, Jiangsu University, Zhenjiang 212013, Chin)

机构地区：[1]江苏大学电气信息工程学院,镇江212013

出　　处：《农业工程》2018年第3期65-69,共5页AGRICULTURAL ENGINEERING

基　　金：国家自然科学基金项目(项目编号:31471413);江苏高校优势学科建设工程项目PAPD(项目编号:苏政办发20116号);江苏省六大人才高峰项目(项目编号:ZBZZ-019);大学生实践创新训练计划项目(项目编号:201710299007Z);大学生科研立项项目(项目编号:16A079)

摘　　要：为了实现无损检测生菜叶片中重金属镉的含量,以高光谱技术为研究手段,研究一种基于高光谱技术的精确、快速和有效检测生菜中重金属镉含量的方法。首先,使用高光谱图像采集系统获取生菜高光谱图像,并提取光谱数据,对提取出的光谱数据采用连续投影算法(SPA)和基于权重回归系数的特征选择算法进行特征提取,建立预测生菜叶片中镉含量的最小二乘支持向量回归(LSSVR)模型。结果表明:SPA-LSSVR模型性能最佳,其中预测集决定系数为0.927 3,均方根误差为0.093 mg/kg。因此,利用高光谱技术结合SPA-LSSVR模型对生菜叶片中重金属镉含量进行预测是可行的,可为实际应用提供技术支持和参考。In order to achieve nondestructive detection of heavy metal cadmium in lettuce leaves,hyperspectral technology was used as research method. A precise,rapid and effective method for detecting heavy metal cadmium in lettuce based on hyperspectral technique was studied. First,hyperspectral image acquisition system was used to obtain hyperspectral image of lettuce,and spectral data was extracted. Extracted spectral data were extracted by successive projections algorithm and weighted regression coefficients,then a LSSVR model for predicting cadmium content in lettuce leaves was established. Results showed that SPA-LSSVR model has the best performance,in which the coefficient of determination set was 0. 927 3 and the root mean square error was 0. 093 mg/kg. Therefore,it is feasible to predict heavy metal cadmium content in lettuce leaves by using hyperspectral technique combined with SPA-LSSVR model,which can provide technical support and reference for practical application.

关 键 词：高光谱图像 重金属 镉 连续投影算法 最小二乘支持向量回归模型 

分 类 号：O657[理学—分析化学]