玉米叶片铜铅胁迫高光谱识别研究  

Hyperspectral Identification of CopperLead Stress in Maize Leaves

作　　者：杨可明[1] 高伟 陈改英[2] 赵恒谦[1] 韩倩倩 李艳茹 YANG Keming;GAO Wei;CHEN Gaiying;ZHAO Hengqian;HAN Qianqian;LI Yanru(College of Geoscience and Surveying Engineering,China University of Mining and Technology(Beijing),Beijing 100083,China;College of Landscape Architecture,Beijing University of Agriculture,Beijing 102206,China)

机构地区：[1]中国矿业大学(北京)地球科学与测绘工程学院,北京100083 [2]北京农学院园林学院,北京102206

出　　处：《农业机械学报》2021年第6期215-222,共8页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金项目(41971401);中央高校基本科研业务费专项资金项目(2020YJSDC02)。

摘　　要：为了区分玉米叶片重金属胁迫种类,提出一种基于高光谱的铜铅胁迫识别方法。分别以叶片0.1~2.0阶分数阶导数(FOD)光谱中红边位置与任意两波长处的光谱值构建玉米叶片的红边铜铅敏感指数(RECLSI)集群,计算各集群中指数与胁迫类型的相关系数,以相关系数最大值、最小值对应的RECLSI构建铜铅识别特征(CLIF),在CLIF的二维分布出现与胁迫类型相关的聚类时建立胁迫识别界限(SIB),从而实现铜铅胁迫识别。研究表明:各RECLSI集群中指数与胁迫类型相关系数的最大值、最小值随FOD光谱阶次的增加分别呈先升后降、先降后升的趋势,其中相关系数最大值、最小值的极点分别出现在1.3、1.4阶FOD光谱对应的RECLSI集群中;0.7~1.5阶FOD光谱的CLIF二维分布呈现出与胁迫类型相关的聚类,根据CLIFSIB能够不同程度地实现铜铅胁迫识别;1.2阶FOD光谱的CLIFSIB识别效果最好,试验集精度为100%,验证集精度为81.25%。基于FOD光谱的CLIFSIB玉米叶片铜铅胁迫识别方法在部分阶次能够获得良好且稳定的识别结果,具有可行性和有效性。A two-dimensional method for the detection of copperlead stress in maize leaves based on hyperspectrum was proposed.Multi-order red-edged copperlead sensitivity index(RECLSI)cluster of maize leaves was constructed by using the spectral values of the red edge position and two wavelengths in the 0.1~2.0 fractional order derivative(FOD)spectrum.The correlation coefficient between index and stress type in each cluster was calculated.The copperlead identification features(CLIF)were constructed with the maximum and minimum correlation coefficients.The stress identification boundary(SIB)was established when clustering related to stress type appeared in the two-dimensional distribution of CLIF,enabling copperlead stress identification.It was found that the maximum and minimum values of the correlation coefficient between the index and the stress type in each RECLSI cluster showed a trend of firstly rising and then falling,or firstly falling and then rising with the increase of FOD spectrum order.The poles appeared in the RECLSI clusters corresponding to the 1.3 order and 1.4 order FOD spectra,respectively.The two-dimensional CLIF distribution of 0.7~1.5 order FOD spectra showed clustering in relation to the type of stress,and the identification of copperlead stresses could achieve different degrees according to CLIFSIB.In the test set,the identification effect of CLIFSIB in 1.2 order FOD spectrum was the best,with the accuracy(A)of 100%,and the 0.9 order,1.0 order and 1.3 order FOD spectra corresponded to value of A of more than 90%.In the verification set,the identification effect of CLIFSIB in 1.4 order FOD spectrum was the best,A was 87.5%,and the A was 81.25%at the 1.2 order FOD spectrum.The CLIFSIB maize leaf copperlead stress discrimination method based on FOD spectrum can effectively discriminate the stress types and it was stable.

关 键 词：玉米叶片 铜铅胁迫 重金属污染 高光谱识别 二维平面 

分 类 号：S127[农业科学—农业基础科学] P575.4[天文地球—矿物学]