| 玉米叶片铜铅胁迫高光谱识别研究 |
| |
| 引用本文： | 杨可明,高伟,陈改英,赵恒谦,韩倩倩,李艳茹.玉米叶片铜铅胁迫高光谱识别研究[J].农业机械学报,2021,52(6):215-222. |
| |
| 作者姓名： | 杨可明 高伟 陈改英 赵恒谦 韩倩倩 李艳茹 |
| |
| 作者单位： | 中国矿业大学(北京);北京农学院 |
| |
| 基金项目： | 国家自然科学基金项目(41971401)和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CLIF-SIB能够不同程度地实现铜铅胁迫识别；1.2阶FOD光谱的CLIF-SIB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CLIF-SIB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。
|
| 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 |
| 收稿时间： | 2020/7/23 0:00:00 |
Hyperspectral Identification of Copper-Lead Stress in Maize Leaves |
| |
| YANG Kemin,GAO Wei,CHEN Gaiying,ZHAO Hengqian,HAN Qianqian,LI Yanru.Hyperspectral Identification of Copper-Lead Stress in Maize Leaves[J].Transactions of the Chinese Society of Agricultural Machinery,2021,52(6):215-222. |
| |
| Authors: | YANG Kemin GAO Wei CHEN Gaiying ZHAO Hengqian HAN Qianqian LI Yanru |
| |
| Institution: | China University of Mining and Technology (Beijing);Beijing University of Agriculture |
| |
| Abstract: | A two-dimensional method for the detection of copper-lead stress in maize leaves based on hyperspectrum was proposed. Multi-order red-edged copper-lead 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 copper-lead 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 copper-lead 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 copper-lead stresses could achieve different degrees according to CLIF-SIB. In the test set, the identification effect of CLIF-SIB 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 CLIF-SIB 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 CLIF-SIB maize leaf copper-lead stress discrimination method based on FOD spectrum can effectively discriminate the stress types and it was stable. |
| |
| Keywords: | maize leaves copper-lead stress heavy metal pollution hyperspectral identification two-dimensional planes |
|
| 点击此处可从《农业机械学报》浏览原始摘要信息 |
| 点击此处可从《农业机械学报》下载免费的PDF全文 |
|
