基于器官造型的植物精细重建

Refine reconstruction of plant organs has great value for plants nondestructive precise measurements, and requires the nowledge and technology such as computer vision and computer graphics, etc. Current methods are mostly for a single plant organ econstruction, and the geometric characteristics are not kept well, which is harmful for the reconstruction of plant organ growth model. This project aims to conduct refine reconstruction on a variety of plant organs. Firstly, the point cloud features of plant organs are analyzed, various registration methods are explored to obtain a unified point cloud data. Then how to segment point clouds into plant organs will be studied. Secondly, multi-scale plant organs models will be constructed and to match the point cloud data. Missing point cloud data will be predicted by integrating various means, and deformation methods will be designed for accurate reconstruction of plant organs. Finally, fast rendering will be implemented. The innovation of the project is combining the plant organs reconstruction with the real point cloud data, combined fusing the.scan data with the image data, and integrating modeling with deformation, in order to obtain strong fidelity, accuracy and usefulness.

植物器官的精细重建对植物的无损和精确测量，有较大的应用价值和学术价值，需要应用计算机视觉和计算机图形学科知识和技术。现有的方法大多针对单一的植物器官进行重建，且重建后的植物器官模型几何特征显著性不足，难以达到植物器官精细重建的水平。本项目旨在研究多种植物器官的精细重建方法。首先分析植物器官点云的特征，研究多种手段获取点云数据的统一注册方法，研究点云数据中植物器官的分割方法。其次，构建多尺度植物器官的可视化模型，并与点云数据匹配，融合多种方法预测缺失的点云数据，设计变形方法精确重建植物器官，最后进行快速渲染。项目的创新之处在于将植物器官重建与真实点云数据分析相结合，扫描数据与图像数据相融合，变形与造型相结合，这些技术具有较强的逼真度、准确性和实用性。

植物器官的精细重建对植物的无损和精确测量，有较大的应用价值和学术价值，项目主要研究植物器官的造型、植物器官的精细重建、植物器官的特征检测与分析、图像与点云数据的基本处理等方面内容。植物器官的造型主要包括树木、叶片、花朵、果实造型；植物器官精细重建主要包括模型与图像相结合、模型与骨架相结合、模型与点云结合进行精细重建。项目的创新之处在于将植物器官重建与真实点云数据分析相结合，点云数据与图像数据相融合，变形与造型相结合。研究成果可服务于精准农林业、虚拟现实与数字娱乐、农林业科研教育等众多领域。本项目在执行过程中出版专著1部，发表论文17篇，其中SCI、EI检索论文10篇，北大中文核心3篇;申请中国发明专利10项，其中已授权4项、通知授权1项、公示5项;获植物造型与重建软件著作权2项;培养硕士生6人。