玉米光谱变分模态分解及其铜铅污染的VMD-CR-ED测度模型
投稿时间:2017-06-26  修订日期:2017-09-05  点此下载全文
引用本文:
摘要点击次数: 220
全文下载次数: 
作者单位E-mail
杨可明 中国矿业大学北京 地球科学与测绘工程学院 ykm69@163.com 
孙彤彤 中国矿业大学北京 地球科学与测绘工程学院 ninastt@163.com 
张伟 中国矿业大学北京 地球科学与测绘工程学院  
程龙 中国矿业大学北京 地球科学与测绘工程学院  
王晓峰 中国矿业大学北京 地球科学与测绘工程学院  
基金项目:国家自然科学基金项目(面上项目,重点项目,重大项目)
中文摘要:农作物重金属污染监测是当今高光谱遥感研究的重要内容之一。重金属污染会引发农作物光谱畸变,但不同污染程度的作物光谱在形态上仍具有极高相似度,所以监测其畸变的关键问题之一是如何提取相似光谱间污染的弱差信息。本研究通过设置不同铜(Cu)、铅(Pb)胁迫梯度下的玉米盆栽实验,在测定了玉米叶片光谱、叶片中Cu2+、Pb2+及叶绿素含量基础上,将变分模态分解(VMD)理论引入到光谱信息弱变化监测中,并结合包络线去除(CR)与欧式距离(ED)方法构建了光谱弱差信息的VMD-CR-ED测度模型,通过与光谱角(SA)与光谱相关系数(SCC)等常规光谱测度算法进行应用比较,证明VMD-CR-ED模型在有效区分玉米叶片的不同胁迫程度方面更具有优越性。同时,将VMD-CR-ED模型应用于玉米叶片光谱的紫光、绿峰、黄边、红边、近红外平台、近谷、近峰A、近峰B等8个特征子波段,测度结果得出:在每个特征子波段内,VMD-CR-ED模型测度的VCE值与叶片中Cu2+和Pb2+含量呈正相关,与叶绿素含量呈负相关;VMD-CR-ED模型应用于紫光、绿峰、黄边、红边、近红外平台和近峰B的6个光谱特征子波段时,VCE值与叶片中Cu2+含量相关性较高(相关系数达到0.99以上),可较好监测Cu2+污染与区分其污染程度;VMD-CR-ED模型应用于近红外平台、近谷、近峰A的3个光谱特征子波段时,VCE值与叶片中Pb2+含量的相关系数高达0.98以上,可利于Pb2+污染监测与污染程度区分。最后,基于检验数据验证了VMD-CR-ED测度模型在农作物铜铅污染监测方面具有可行性和可靠性。
中文关键词:光谱分析  变分模态分解  农作物  铜铅重金属  污染监测模型
 
Spectral Variational Mode Decomposition and VMD-CR-ED Model on Measuring Copper and Lead Pollution of Corn
Abstract:The monitoring of heavy metal pollution in crops is one important point of hyperspectral remote sensing study. Heavy metal pollution can cause spectral distortion of crop, but the crop spectra of different pollution degrees still has very high similarity in shape, so one of the key problems in monitoring its distortion is how to extract the weak difference information between similar spectra. In this study, the Cu2+, Pb2+ and chlorophyll content in corn leaf was measured by setting the pot experiment of corn in different copper (Cu) and lead (Pb) stress gradient, a VMD-CR-ED measurement model was established by introducing the variational mode decomposition (VMD) theory into weak change monitoring of spectral information and combining with the continuum removal (CR) and euclidean distance (ED) methods, the model was proved to be superior in distinguishing the different pollution degrees of corn leaves by compared with SA and SCC. At the same time, the VMD-CR-ED model was applied to eight characteristic sub-bands of purple, green peek, yellow edge, red edge, near-infrared platform, near-valley, near-peak-A and near-peak-B, the results showed that the VCE values of VMD-CR-ED model were positively correlated with the contents of Cu2+ and Pb2+ in leaves, and negatively correlated with chlorophyll content in each characteristic sub-band. When the VMD-CR-ED model was applied to the six spectral characteristic sub-bands of purple, green peek, yellow edge, red edge, near-infrared platform, and near-peak-B, the correlation coefficient between the VCE and the Cu2+ contents in leaves were higher (the correlation coefficient were above 0.99), which can be better monitor the pollution and distinguish the pollution degree of Cu2+; When the VMD-CR-ED model is applied to the three spectral characteristic sub-bands of near-infrared platform, near-valley and near-peak-A, the correlation coefficient between VCE value and Pb2+ contents in the leaves were above 0.98, which is also beneficial to monitor the pollution and distinguish the pollution degree of Pb2+. Finally, based on the test data, it is proved that the VMD-CR-ED measurement model is feasible and reliable in the monitoring of copper and lead pollution of crops.
keywords:Spectral  analysis, Variational  Mode Decomposition, Crop, Copper  and lead  heavy metal, Pollution  monitoring model
查看全文  查看/发表评论  下载PDF阅读器

版权所有:《红外与毫米波学报》编辑部