| 基于无人机可见光图像估算甘肃鼢鼠对马铃薯的为害量 |
| 点此下载全文 |
| 引用本文：甄磊,郭永旺,秦萌,王文慧,王登.基于无人机可见光图像估算甘肃鼢鼠对马铃薯的为害量.植物保护学报,2022,49(6):1697-1704 |
| DOI：10.13802/j.cnki.zwbhxb.2022.2022051 |
| 摘要点击次数: |
| 全文下载次数: |
|
|
| 中文摘要:为估算农田害鼠对作物的为害损失量，使用无人机拍摄甘肃鼢鼠Eospalax cansus为害的22块马铃薯样地正射影像图，首先目视解译标定各样地为害区域，计算为害率，并据此划分各样地鼠害为害等级；随后运用基于规则的特征提取法和监督分类法（支持向量机分类法和神经网络分类法）对各样地裸地和植被进行分类，结合对照区裸地率计算各样地的鼠害为害裸地率；通过构建鼠害为害裸地率与马铃薯产量的线性关系模型来评估不同分类法获得的鼠害为害裸地率的精确性；用拟合度最好的线性关系模型估算无鼠害及当前鼠害水平下的马铃薯产量，最终计算全部样地鼠害造成的马铃薯损失量。结果表明，基于规则的特征提取法、支持向量机分类法和神经网络分类法的地物分类精度分别为71.46%、99.33%和98.84%，3种分类方法获得的样地鼠害为害裸地率与马铃薯实际产量均呈显著线性相关，但神经网络分类获得的结果拟合度最好，R2为0.558。利用该方法估算的甘肃鼢鼠造成的马铃薯产量损失量为7 032.75 kg/hm2。 |
| 中文关键词:甘肃鼢鼠 马铃薯 无人机遥感 可见光图像 为害量 |
| |
| Assessment of potato damages caused by Gansu zokor Eospalax cansus using RGB images from unmanned aerial vehicle |
| Author Name | Affiliation | E-mail | | Zhen Lei | College of Grassland Science and Technology, China Agricultural University, Beijing 100193, China | | | Guo Yongwang | National Agro-Tech Extension and Service Center, Beijing 100026, China | | | Qin Meng | National Agro-Tech Extension and Service Center, Beijing 100026, China | | | Wang Wenhui | Dingxi Plant Protectionand Quarantine Station, Dingxi 743000, Gansu Province, China | | | Wang Deng | College of Grassland Science and Technology, China Agricultural University, Beijing 100193, China | wangdeng@cau.edu.cn |
|
|
| Abstract:In order to quantitatively evaluate the crop yield loss caused by rodents, orthophoto images of 22 potato sample plots damaged by Gansu zokor Eospalax cansus were taken using unmanned aerial vehicle (UAV) in the study site of Dingxi City, Gansu Province. Based on these images, the damaged areas in each sample plot were firstly identified by visual interpretation, the damage rate was calculated, and then the damage levels of each sample plot were classified according to the damage rate. The rulebased feature extraction method and supervised classification method (support vector machine classification algorithm and neural network classification) were then applied to classify the bare land and vegetation, and the rate of bare land caused by rodents were calculated using the results of bare land and vegetation combined with the bare land rate of damage-free areas in the sample plots. Finally, the yields of eight potato sample plots with different damage levels were measured, and simple linear models of the rate of bare land caused by rodents and potato yield were constructed to assess the accuracy of the rate. The results showed that the rule-based feature extraction method for feature classification, the support vector machine classification method and the neural network classification method had an accuracy of 71.46%, 99.33% and 98.84%, respectively. The rate of bare land caused by rodents in the damaged sample plots obtained from each classification method was significantly correlated with the actual yield, but the classification result of the neural network showed the best fit (R2=0.558), and a potato loss of 7 032.75 kg/hm2 caused by Gansu zokor was calculated using the linear model. |
| keywords:Eospalax cansus potato UAV remote sensing RGB images damage |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
|
您是本站第 8734585 版权所有：植物保护学报 京ICP备05006550号-2 主管单位：中国科协 主办单位：中国植物保护学会、中国农业大学 地址：北京市圆明园西路2号 中国农业大学植物保护学院 植物保护学报编辑部 电话：010-62732528 电子邮件：zbxb@cau.edu.cn 技术支持： 北京勤云科技发展有限公司 |
