【作者】 张昭；

【导师】 方慧；

【作者基本信息】 浙江大学 ， 农业机械化工程， 2016， 硕士

【摘要】 研究植物三维信息的获取可以用于监控作物生长过程中的动态数据,为研究植物生长机理提出新的手段与思路,是发展精细农业的重要内容。目前,对于植物三维模型的研究处于起步阶段,对于植物点云模型的深入研究并不多,与植物生理信息结合的较少。本文借助Point Cloud Library(PCL)开发库,对于植物点云模型进行深入研究,主要研究工作和研究成果包括以下几个方面。(1)针对实验室内三维扫描仪无法通过修改参数减小误差的情况,介绍了一种基于标准平板的误差分析方法,并根据分析结果提出两种误差补偿算法,最后对误差补偿算法的效果进行了评估。(2)针对植物叶片中存在的病斑提取问题,提出了基于颜色信息的区域生长算法,实现了对植物叶片病斑的区域分割,并取得了较好的效果。同时讨论了该算法中各个参数对分割结果的影响。最后对算法的分割效果进行评估。(3)由于高光谱数据在研究植物生理、病斑检测等问题中的作用巨大,提出了一种将高光谱数据与植物点云模型相结合的算法。对比常用的图像匹配算法,分析了不同算法对最终结果的影响。通过本研究,实现了高光谱数据与点云模型的相结合,可以为解决病斑检测、成分分析等问题提供了一个新的思路。更多还原

【Abstract】 Study of 3D plant information, as an important part of precision agriculture, can obtain dynamic data in the process of plant growth which offers new ideas for the mechanism study. At present, research of 3D plant models is in its infancy, and a few researches has combined with plant physiology information.We have depth research about 3D plant models with the Point Cloud Library in this paper.This paper includes:(1) In view of the 3D scanner’s parameter can’t be modified to reduce errors in our lab, we introduced an error analysis method based on the standard board. We proposed two error compensation algorithms based on the result of error analysis, and evaluate the results of error compensation.(2) In view of the extraction lesion for the 3D leaf models, we proposed a region growth algorithm based on the color difference. We have gotten good results in the field of extraction lesion for the 3D leaf models. In addition, we discussed the influence of 4 parameters in the algorithm, and evaluate the results of extraction lesion.(3) Hyperspectral data has been widely used throughout the study of plant physiology and quality inspection. We proposed an algorithm which combined hyperspectral data with 3D leaf models. We compared different image matching algorithms and evaluate the results.Through this research, We achieve the combination of hyperspectral data and point cloud model, the combination can provide new ideas for the problems of lesion detection and composition analysis.更多还原