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  

Method for rapidly acquiring images of field-grown crops using aquad-rotor UAV and its application

作　　者：李晓鹏 胡鹏程[1] 徐照丽[2] 晋艳[2] 杨宇虹[2] 郑邦友 段涛[1] 郭焱[1] 

机构地区：[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6500212 [3]CSIRO Agriculture and Food,Queensland Biosciences Precinct

出　　处：《中国农业大学学报》2017年第12期131-137,共7页Journal of China Agricultural University

基　　金：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)

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20m)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。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 3 Dreconstruction.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.

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 

分 类 号：V279.2[航空宇航科学与技术—飞行器设计]