| 利用无人机图像颜色与纹理特征数据在小麦生育前期对产量进行预测 |
| |
| 引用本文： | 刘欣谊,仲晓春,陈晨,刘涛,孙成明,李冬双,刘升平,王建军,丁大伟,霍中洋.利用无人机图像颜色与纹理特征数据在小麦生育前期对产量进行预测[J].麦类作物学报,2020,40(8):1002-1007. |
| |
| 作者姓名： | 刘欣谊 仲晓春 陈晨 刘涛 孙成明 李冬双 刘升平 王建军 丁大伟 霍中洋 |
| |
| 作者单位： | 江苏省作物遗传生理重点实验室/江苏省作物栽培生理重点实验室,扬州大学农学院,江苏扬州225009;江苏省粮食作物现代产业技术协同创新中心,扬州大学,江苏扬州225009;中国农业科学院农业信息化研究所,北京100081;张家港市农业试验站,江苏张家港215616 |
| |
| 基金项目： | 国家重点研发计划项目（2018YFD0300805）；国家自然科学基金项目（31671615，31701355，31872852）；江苏高校优势学科建设工程资助项目(PAPD)；国家博士后基金（2016M600448,2018T110560）；苏州市农业科技创新项目（SNG2017064） |
| |
| 摘 要： | 为了实现基于无人机的小麦产量快速预测，通过不同种植密度、氮肥和品种的田间试验，应用无人机航拍获取小麦生育前期（越冬前期和拔节期）的RGB图像，通过图像处理获取小麦田间颜色和纹理特征指数，并在小麦收获后测定实际产量。通过分析不同颜色和纹理特征指数与小麦产量的关系，筛选出适合小麦产量预测的颜色和纹理特征指数，建立小麦产量预测模型并进行验证。结果表明，小麦生育前期图像颜色指数与产量的相关性较好，而纹理特征指数相关性较差。对越冬前期利用单一颜色指数NDI构建的产量预测模型验证时，R为0.541，RMSE为671.26 kg·hm-2；对拔节期用单一颜色指数VARI构建的产量预测模型验证时，R为0.603，RMSE为639.78 kg·hm-2，预测结果比较理想，但不是最优。对越冬前期颜色指数NDI和纹理特征指数ENT相结合构建的产量预测模型验证时，R和RMSE分别为0.629和611.82 kg·hm-2，比单一颜色指数模型分别提升16.27%和减小8.85%；对拔节期颜色指数VARI和纹理特征指数COR相结合构建的产量预测模型验证时，R和RMSE分别为0.746和510.29 kg·hm-2，较单一颜色指数模型分别提升23.71%和减小20.24%。上述结果说明，将无人机图像颜色和纹理特征指数相结合建立的估产模型精度较高，可在小麦生育前期对产量进行有效预测。
|
| 关 键 词： | 小麦 无人机图像 颜色指数 纹理特征指数 生育前期 产量预测 |
Prediction of Wheat Yield Using Color and Texture Feature Data of UAV Image at Early Growth Stage |
| |
| LIU Xinyi,ZHONG Xiaochun,CHEN Chen,LIU Tao,SUN Chengming,LI Dongshuang,LIU Shengping,WANG Jianjun,DING Dawei,HUO Zhongyang.Prediction of Wheat Yield Using Color and Texture Feature Data of UAV Image at Early Growth Stage[J].Journal of Triticeae Crops,2020,40(8):1002-1007. |
| |
| Authors: | LIU Xinyi ZHONG Xiaochun CHEN Chen LIU Tao SUN Chengming LI Dongshuang LIU Shengping WANG Jianjun DING Dawei HUO Zhongyang |
| |
| Abstract: | In order to achieve the rapid prediction for wheat yield based on unmanned aerial vehicle (UAV),the RGB images were obtained using UAV at early growth stage (pre-winter stage and jointing stage) of wheat through the field experiments with different density,nitrogen and variety in this study. The color and texture feature indices of wheat field were obtained through image processing,and the actual yield was determined after wheat harvest. The relationship between the different color and texture feature indices and wheat yield was analysed,and the suitable color and texture feature indices for wheat production prediction were selected to establish wheat yield prediction model. The results showed that the correlation between image color index and yield at the early stage of wheat growth was good,while the correlation of texture feature index was poor. At the pre-wintering stage,the yield prediction model constructed using a single color index NDI was verified to have R of 0.541 and RMSE of 671.26 kg·hm-2. At jointing stage,the yield prediction model constructed by using a single color index VARI was verified to have R of 0.603 and RMSE of 639.78 kg·hm-2. The result was more ideal,but it was not the optimal result. The yield prediction model constructed by combining color index NDI and texture feature index ENT at pre-wintering stage was verified to have R of 0.629,and the value of R was promoted by 16.27% comparing to that of the single color index model. And RMSE was 611.82 kg·hm-2,decreased by 8.85% comparing to that of the single color index model. The yield prediction model constructed by combining color index VARI and texture feature index COR at jointing stage was verified to have R of 0.746,and the value of R was promoted by 23.71% comparing to that of the single color index model. And RMSE was 510.29 kg·hm-2,decreased by 20.24% comparing to that of the single color index model. The above results showed that the yield estimation model established by combining the color and texture feature indices of UAV images was more accurate and could effectively predict the wheat yield at the early stage of wheat growth. |
| |
| Keywords: | Wheat UAV image Color index Texture feature index Early growth stage Yield prediction |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《麦类作物学报》浏览原始摘要信息 |
| 点击此处可从《麦类作物学报》下载免费的PDF全文 |
|
