激光点云和多视影像融合的油菜植株重建研究

项目名称： 激光点云和多视影像融合的油菜植株重建研究

项目编号： No.41301522

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 天文学、地球科学

项目作者： 林承达

作者单位： 华中农业大学

项目金额： 25万元

中文摘要： 油菜是农业生产的重要对象，其形态数据和三维模型在生长监测、形态分析、遗传育种等方面有重要的应用价值，如何快速地获取三维形态信息，生成高质量的三维模型成为农业信息化发展的一个新趋势。激光扫描技术为目标重建提供了一种新的途径，融合激光点云和影像数据则为复杂目标的三维重建研究带来了新的机遇。本项目以苗期油菜地上植株部分为研究对象，通过融合高精度激光扫描数据和多视高分辨率影像数据，实现准确、高精度的地上植株重建。主要研究内容包括：（1）基于同步影像的激光扫描系统和数码相机系统间的配准；（2）基于共面条件约束的两视点云拼接和多条件共同约束的多视点云无缝拼接；（3）融合点云和影像数据的目标细部特征重建。本项目研究成果可以提供实时的油菜形态重建，为农学研究者提供便捷的油菜可视化方法，从而服务于油菜的理想株型研究、栽培育种等。

中文关键词： 激光扫描；三维重建；点云配准与拼接；数据融合；

英文摘要： Rape is one the most important objects in agricultural production. The morphological data and three dimensional models of plants can be applied in many applications, such as growth detection, morphological analysis, genetic breeding and others. The emergence of terrestrial laser scanner has contributed to three dimensional reconstruction of objects as an alternative fasion. The integration of laser point clouds and digital images has proivided a promising way for complex object reconstruction. This research presents a new methodology for rape reconstruction by integrating point clouds and digital images. The reserch include three parts: (1) Co-registration of laser scanner and opticla digital camera based on synchronized image and high resolution digital image, (2)two view point cloud registration based on coplanarity constraints and Iterative Closest Point algorithm, and multiview point cloud registration constrained by coplanarity and closed condition, (3) Detailed feature reconstruction by integrating point clouds and digital image data. The objective of the research is to realize three dimensional reconstruction and visualization of rape plant, so as to serve agricultural production.

英文关键词： ground laser scanner；3D reconstruction；point cloud registration；intergration of point cloud and digital image；