摘要
针对玉米植株三维骨架的快速获取问题,提出了一种基于Kinect自动获取玉米植株三维骨架的方法,采用由Kinect转换的点云数据,利用阈值分割的方法删除背景数据,根据玉米植株三维点云模型的高度分层,对每层点组采用二次循环迭代方法聚类,对每层中各子类求取中心点。中心点的集合即认为是玉米植株的骨架点。采用二次B样条曲线插值拟合连线。实验结果表明:方法获得的玉米植株三维线性骨架与通过三维数字化仪获得的三维线性骨架一致性较好,能够满足实验要求。
Aiming at acquiring three-dimensional skeleton of maize plant efficiently, an automatic three-dimensional maize plant skeleton extraction method based on Kinect data was proposed. Kinect data was transformed into point cloud data. Background point cloud was deleted by threshold segmentation and the point cloud was stratified according to the height of the 3D model. Each layer was clustered by two circle iteration and the center points of each cluster could be obtained. The set of center points could be considered as the skeleton points of maize. The quadratic B-spline interpolation curve was used to obtain relatively accurate and smooth linear skeleton of maize plant. Experimental result shows that the 3D maize plant linear skeleton obtained by this method has a good consistency of 3D digitizer skeleton and can meet the requirements of rapid experimental data acquirement.
作者
方志力
温维亮
郭新宇
吴建伟
Fang Zhili Wen Weiliang Guo Xinyu Wu Jianwei(Beijing Research Center for Information Technology in Agriculture/Beijing Key Lab of Digital Plant/National Engineering Research Center for Information Technology in Agriculture/Key Laboratory of Agri-information Ministry of Agriculture, P R China/Beijing Agriculture lnternet of Things Engineering Technology Research Center, Beijing 100097, China Shanghai Ocean University Information College, Shanghai 201306, China Beijing PAIDE Science and Technology Development Co. Ltd., Beijing 100097, China)
出处
《系统仿真学报》
CAS
CSCD
北大核心
2017年第3期524-530,共7页
Journal of System Simulation
基金
国家自然科学基金(31171454)
国家863计划(2013AA102404)
北京市科技计划(D151100004215004
Z151100003215008)
