基于点云的果树冠层叶片重建方法

发布时间：2018-06-03 06:14

本文选题：激光仪 + 聚类算法　；参考：《农业工程学报》2017年S1期

【摘要】：精确的果树三维冠层结构是农业科研人员进行功能结构模型研究的重要载体,该文提出一种快速、精确、自动的果树冠层叶片重建方法。首先根据带叶果树点云的局部和全局特征,建立椭球分层的点云密度收缩方法实现器官点云分离,然后利用邻近传播主成分分析算法实现叶片特征参数的求解,利用Laplacian收缩算法实现冠层骨架点的连通,从而实现冠层叶片的快速自动重建。最后利用C++及Point Cloud Library(PCL)点云库,开发果树叶片点云冠层自动重建系统,对苹果树、柑橘树等不同类型果树进行算法验证,结果表明该方法能够正确识别出的叶片数占冠层总叶片数的90%以上,叶面积指数的正确率大于95%,叶片倾角偏离5?以内的叶片数占总叶片数的90%以上。该方法得到了较好的可视化效果和叶冠三维重建精度,可为后期树体冠层内光合作用的研究、整形修剪、农业仿真试验等提供参考。
[Abstract]:The accurate three-dimensional canopy structure of fruit trees is an important carrier for agricultural researchers to study the functional structure model. In this paper, a fast, accurate and automatic method for canopy leaf reconstruction of fruit trees is proposed. Firstly, according to the local and global characteristics of point cloud in fruit trees with leaves, an ellipsoidal and stratified point cloud density contraction method is established to separate organs from point clouds, and then the method of neighborhood propagation principal component analysis is used to solve the characteristic parameters of leaves. The Laplacian shrinkage algorithm is used to realize the connectivity of the canopy skeleton, so that the canopy blade can be reconstructed automatically. Finally, using C and Point Cloud Library, an automatic canopy reconstruction system is developed to verify the algorithm of apple trees, citrus trees and other different types of fruit trees, such as apple trees, citrus trees and other different types of fruit trees. The results show that the number of leaves can be correctly identified by this method, which accounts for more than 90% of the total leaf number in the canopy, and the accuracy of leaf area index is greater than 95%, and the angle of leaf inclination deviates from 5%. The number of leaves within the total number of leaves accounted for more than 90%. This method has better visualization effect and 3D reconstruction accuracy of leaf crown, which can be used as reference for the study of photosynthesis in canopy layer, plastic pruning and agricultural simulation experiment.
【作者单位】：北京林业大学信息学院;国家农业信息化工程技术研究中心;北京农业信息技术研究中心;数字植物北京重点实验室;
【基金】：国家“863”高技术研究发展计划课题(2013AA102405) 北京市科技新星资助项目(Z131101000413022) 北京市农林科学院科技创新团队(JNKYT201604)联合资助
【分类号】：S66;TP391.41

【参考文献】