基于地基激光雷达数据的单株玉米三维建模

发布时间：2018-08-17 14:46

【摘要】：为探索地基激光雷达数据在植物真实三维建模上的应用,研究基于MeshLab软件的三维Alpha-Shape算法和移动立方体算法进行玉米植株三维真实模型的重建。三维Alpha-Shape算法基于离散激光雷达点云构建Delaunay三角网,通过Alpha值的设置判断各单纯形是否保留,完成玉米植株三维重建。移动立方体算法则通过查找每个体素与等值面相交的交面,所有交面连在一起即为所求的等值面。Alpha值分别设置为0.006 290、0.012 638、0.018 635,采用Alpha-Shape算法进行单株玉米三维重建,试验结果表明:当Alpha值设置较小时,对玉米叶片、茎秆等细节信息表达的比较精细,但许多叶片中间不连续;Alpha值设置较大时,对玉米植株细节信息表达不精细。基于移动立方体算法重建出的玉米植株曲面较为光滑,且孔洞较少,但重建结果不完整。
[Abstract]:In order to explore the application of ground-based lidar data in plant real 3D modeling, 3D Alpha-Shape algorithm and moving cube algorithm based on MeshLab software were studied to reconstruct the 3D real model of maize plant. Three-dimensional Alpha-Shape algorithm is based on discrete lidar point cloud to construct Delaunay triangulation network. Through the setting of Alpha value to determine whether each simplex is reserved or not to complete the three-dimensional reconstruction of maize plant. By searching for the intersection between each voxel and the equivalent surface, the moving cube algorithm sets the equivalent surface. Alpha value of all the intersecting surfaces together is 0.006 ~ (290) ~ 0.012 ~ (638) ~ 0.018 ~ (635) respectively. The Alpha-Shape algorithm is used to reconstruct a single plant of maize. The results showed that when the Alpha value was small, the detailed information of maize leaves and stems was fine, but when the Alpha values of many leaves were larger, the detailed information of maize plants was not well expressed. The surface of maize plant reconstructed by moving cube algorithm is smooth and has fewer holes, but the reconstruction result is incomplete.
【作者单位】：中国农业大学信息与电气工程学院/农业部农业信息获取技术重点实验室;北京航天勘察设计研究院有限公司;中国矿业大学(北京)地球科学与测绘工程学院;
【基金】：中央高校基本科研业务费专项资金项目(2013QJ055) 北京高校青年英才计划项目资助(YETP0316)
【分类号】：S513;TN958.98

【参考文献】