| 本文已被：浏览 981次 下载 327次 |
码上扫一扫！ |
| 基于四旋翼无人机快速获取大田植株图像的方法及其应用 |
|
李晓鹏1, 胡鹏程1, 徐照丽2, 晋艳2, 杨宇虹2, 郑邦友3, 段涛1, 郭焱1
|
|
|
| (1.中国农业大学 资源与环境学院, 北京 100193;2.云南省烟草农业科学研究院, 昆明 650021;3.CSIRO Agriculture and Food, Queensland Biosciences Precinct, Brisbane, St Lucia 4067 QLD, Australia) |
|
| 摘要: |
| 利用低空无人机获取农田信息，具有实时以及灵活性高、成本低等优势。为快速、精确监测大田规模化种植作物的生长发育状况，以四旋翼无人机为平台，结合数字图像技术，建立快速获取大田烟株中前期图像的方法。结果表明，在天空辐射条件较稳定的条件下，采用较低的飞行高度（如20 m）航拍获取田块图像，能够得到清晰的拼接图像和三维重建效果；采用基于决策树的植被分割算法将烟草和非植被部分分割后，得到较高精度的大田植株图像。在此基础上进行大田烟草缺苗数估测，所估算的缺苗数与实测值吻合较好。 |
| 关键词: 无人机 植物表型 生长发育 图像拼接 三维点云 植被分割 烟草 |
| DOI：10.11841/j.issn.1007-4333.2017.12.16 |
| 投稿时间：2016-10-07 |
| 基金项目:中国烟草总公司云南省公司科技项目（2017YN07）；云南中烟工业有限责任公司项目（2014YL01） |
|
| Method for rapidly acquiring images of field-grown crops using a quad-rotor UAV and its application |
|
LI Xiaopeng1, HU Pengcheng1, XU Zhaoli2, JIN Yan2, YANG Yuhong2, ZHENG Bangyou3, DUAN Tao1, GUO Yan1
|
| (1.College of Resources and Environmental Sciences, China Agricultural University, Beijing 100193, China;2.Yunnan Academy of Tobacco Agricultural Sciences, Kunming 650021, China;3.CSIRO Agriculture and Food, Queensland Biosciences Precinct, Brisbane, St Lucia 4067 QLD, Australia) |
| Abstract: |
| Because of the characteristics of high flexibility and low cost,there are lots of advantages to use unmanned aerial vehicle(UAV)to collect the real-time information of field.In order to quickly monitoring the growth and development of crops grown in large scale field,a method for rapidly acquiring images of field-grown tobacco using quad-rotor UAV and digital image processing techniques was explored.The results show that under a relatively stable sky radiation condition,the images acquired with low flight height(e.g.20 m) can be used to produce rather high quality ortho-mosaic and 3D reconstruction.The images of individual field-grown tobacco plants were acquired using the vegetation segmentation algorithm based on decision tree to segment tobacco and non-vegetation part.The missing plant number of tobacco in individual plots was computed based on this algorithm and the results agree to the measured values well.Accordingly,this study provided a reliable way for rapidly and accurately accessing the growth and development of field-grown tobacco and studying plant phenotype in large scale. |
| Key words: unmanned aerial vehicle crop phenotyping growth and development ortho-mosaic 3D point cloud vegetation segmentation tobacco |
