摘要
株高和茎粗是玉米株型中重要的表型参数,可以反映玉米的生长状况和活力,地基激光雷达可实现表型参数的快速自动测量。首先,使用地基激光雷达获取田间2个玉米品种在4个生长期的三维点云数据;其次,使用点云处理软件对采集的玉米点云数据进行匹配、分割,人工点云测量玉米表型;之后,对玉米点云进行下采样、直通滤波、基于随机采样一致性平面分割、统计滤波和圆柱分割等处理,分割单株玉米、去除地面点云、滤除离群点、植株点云分层和提取待测量茎秆点云;运用最高点提取和地面点分割,测量玉米株高,椭圆拟合测量茎横截面长轴和短轴长度。结果表明,与田间测量值相比,京农科728人工点云测量株高、茎横截面长轴和短轴长度的均方根误差分别为21.5、1.24、1.86 mm,农大84分别为23.6、1.56、1.23 mm。与人工点云测量值相比,京农科728自动点云测量值的均方根误差分别为10.2、6.65、3.45 mm,农大84分别为7.1、4.95、3.26 mm。研究表明,地基激光雷达点云数据测量株高、茎粗的方法,适用于不同生长期、不同品种的玉米,与人工田间测量方法具有高度的一致性,可以替代人工测量。
Plant height and stem thickness are important phenotypic parameters in the plant type of maize,which can reflect the growth status and vigor of maize.Terrestrial laser scanning can realize the rapid and automatic measurement of phenotypic parameters.Firstly,the three-dimensional point cloud data of two varieties of maize in four growth periods were obtained by using terrestrial laser scanning.Secondly,the point cloud processing software was used to match and segment the collected maize point cloud data,artificial measurement of maize phenotype in point cloud.Then,the maize point cloud was processed by down sampling,through filtering,plane segmentation algorithm based on random sampling consistency,statistical filtering and cylinder segmentation.The results showed that the method can segment individual maize,remove ground point cloud,filter outliers,layer plant point cloud and extract stem point cloud to be measured.The highest point extraction and ground point segmentation were used to measure maize plant height,and the long axis and short axis of stem thickness were measured by ellipse fitting.Compared the artificial point cloud measurement value with the artificial field measurement value,the root mean square error(RMSE)of Jingnongke 728 plant height,long axis and short axis were 2.15 cm,1.24 mm and 1.86 mm,respectively.The RMSE of Nongda 84 plant height,long axis and short axis were 2.36 cm,1.56 mm and 1.23 mm,respectively.Compared the automatic point cloud measurement value with the artificial point cloud measurement value,the RMSE of Jingnongke 728 plant height,long axis and short axis were 1.02 cm,6.65 mm and 3.45 mm,respectively.The RMSE of Nongda 84 plant height,long axis and short axis were 0.71 cm,4.95 mm and 3.26 mm,respectively.The results showed that the method of measuring plant height and stem thickness with terrestrial laser scanning point cloud data can be widely used in different maize varieties with different growth periods.The result of this method was highly consistent with that of the artificial measurement method and it can replace the artificial measurement method.It can provide a fast,automatic and accurate measurement scheme for breeders and crop phenotypists.
作者
苗艳龙
彭程
高阳
仇瑞承
李寒
张漫
MIAO Yanlong;PENG Cheng;GAO Yang;QIU Ruicheng;LI Han;ZHANG Man(Key Laboratory of Modern Precision Agriculture System Integration Research,Ministry of Education,China Agricultural University,Beijing 100083,China;Key Laboratory of Agricultural Information Acquisition Technology,Ministry of Agriculture and Rural Affairs,China Agricultural University,Beijing 100083,China)
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2021年第S01期43-50,共8页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家重点研发计划项目(2019YFE0125500)
中国农业大学建设世界一流大学(学科)和特色发展引导专项资金项目(2021AC006)
中国农业大学研究生自主创新研究基金项目(2021TC111)
关键词
玉米
株高
茎粗
地基激光雷达
点云
maize
plant height
stem thickness
terrestrial laser scanning
point cloud
