Abstract: Ear is maize yield composition organ. A geometric modeling method for maize ear based on three-dimensional (3D) point cloud was proposed to reconstruct high precision ear geometric models. According to the morphological characteristics of maize ear, a handheld 3D scanner Artec Spider was selected to build a fast point cloud acquisition system for maize ear. Point clouds of ear were acquired by the system and were processed by registration, re-sampling and hole repairing operation to get high quality point clouds. Finally, the ear mesh model was generated by  Voronoi graph based method. Results showed that the reconstructed ear models had high fidelity, thus the accuracy was greatly improved comparing with the computer vision based models. This method has an important role in promoting maize ear germplasm resource preservation, ear analyzing in 3D way and construction of 3D template repository of maize organs.

Key words: maize, ear, three-dimensional reconstruction, point cloud

摘要： 果穗是玉米产量的构成器官,为构建高精度玉米果穗几何模型，提出了一种基于三维点云数据的玉米果穗几何建模方法。针对玉米果穗形态结构特征，选取Artec Spider三维扫描仪搭建玉米果穗点云数据快速获取系统并获取果穗点云，然后通过点云配准、重采样和孔洞修补操作得到高质量果穗三维点云，最后基于Voronoi图的网格重建方法重构果穗网格模型。结果表明，所重建的玉米果穗具有较高的真实感，且与基于计算机视觉算法相比精度大幅提高。基于三维点云的玉米果穗几何建模对于玉米果穗的种质资源保存、基于三维数据的果穗考种、玉米器官三维模板资源库构建等工作具有重要的推动作用。

关键词: 玉米, 果穗, 三维重建, 点云