基于三维点云的木荷幼苗表型参数自动测量方法
Automated measurement method of Schima superba phenotype seeding phenotypic parameters based on 3D point cloud
王斐 1周扬 1龙伟 2王斌 2周志春 2吴统贵2
作者信息
- 1. 浙江科技大学信息与电子工程学院,浙江杭州 310023;浙江科技大学中德工程师学院,浙江杭州 310023
- 2. 中国林业科学研究院亚热带林业研究所浙江省林木育种技术研究重点实验室,浙江杭州 311400
- 折叠
摘要
本文提出了一种基于Azure Kinect传感器无损测量三维点云的表型参数测量方法.该方法包括预处理、茎叶分割和表型参数计算3 个步骤.首先通过预处理将植株点云从场景点云中提取出来,在茎叶分割步骤历经骨架化、骨架修剪、茎线识别和叶片分割几个分步骤将木荷植株的茎干和叶片分离,最后得到株高、茎长、茎的方向、叶长和叶角等表型参数.实验结果表明,每个参数的决定系数(R2＞0.85)和均方根误差(RMSE)均达到了精度要求,说明了该方法是稳健和准确的.
Abstract
A method for non-destructively measuring three-dimensional(3D)point cloud phenotypic parameters hased on Azure Kinect sensor is proposed.The method includes three steps:pretreatment,stem leaf segmentation and phenotypic parameter calculation.Firstly,the plant point cloud is extracted from the scene point cloud by preprocessing.In the stem and leaf segmentation step,the stem and leaf of Schima.superba are separated by several steps:skeletonization,skeleton pruning,stem line recognition and leaf segmentation.Finally,the phenotypic parameters such as plant height,stem length,stem direction,leaf length and leaf angle are obtained.The experimental results show that the determination coefficient(R2＞0.85)and root mean square error(RMSE)of each parameter meet the precision requirements.It shows that the method is robust and accurate.
关键词
木荷/Azure/Kinect/点云/骨架化/表型参数提取Key words
Schima superba/Azure Kinect/point cloud/skeletonization/phenotypic parameters extraction引用本文复制引用
基金项目
浙江省"尖兵领雁"研发攻关计划(2021C02038)
浙江省"三农九方"科技写作计划"揭榜挂帅"项目(2023SNJF027)
出版年
2024
NETL
NSTL
万方数据