基于有限细节的植物叶片多密度点云重建算法

Multiple Density Leaf Reconstruction Based on Limited Details

针对粗糙点云在植物模型重建过程中遇到的噪点多、边缘粘合等问题,提出一种基于有限细节的多密度点云重建算法。首先利用Kinect采集到的深度和颜色信息提取出植物叶片点云,并通过颜色信息对原始点云进行稀疏处理,分离开粘合部位,得到理想的点云;然后基于人眼视觉识别的局限性提出了一种有限细节多密度点云重建算法,与传统的网格重建不同,其以点代面通过不断细化点的密度来产生视觉误差上的模糊曲面。实验证明,所提算法的重建效果和速度在一定程度上优于网格重建的。

According to the rough point cloud encountering the problem of noise and edged bonding in the process of plants modeling,we presented a multiple density point cloud reconstruction algorithm based on limited details. Firstly, we used the color information to spare the original points cloud extracted by Kinect, and then separate the adhesive parts to gain ideal point cloud. Simultaneously,a new method named multiple density with limited details reconstructive algorithm are provided on the basis of limitation of human recognition. The algorithm is different from traditional grid reconstruction, it generates the fuzzy surface by refining the density of points in order to reach the purpose of surface. Finally, the experiments show that the result and speed of the new algorithm are better than mesh reconstruction in a certain extent.