摘要
提出一种基于最小割的彩色点云分割算法,首先找到彩色点云中每个点的临近点,然后与父点连接,再找到距离最近的2个块并连接,重复连接距离最小的2个块,直到只剩1个块,从而生成1张点云图。根据2点之间的欧氏距离和颜色空间距离设置2点连线的权值,当选择待分割点后根据设定的阈值大小和分割范围,查找被选点分割范围内的所有符合条件的点,然后通过区域生长算法对符合条件的点进行生长,进而得到颜色相近的三维连通区域。实验证明所提方法可行,可有效解决三维分割中连通性的问题。
Most of the segmentation methods of point cloud are based on the two-dimensional image processing which geometric information has been ignored. This leads to non-connected regions with similar color that would be easily divided in the same area. The paper proposes a segmentation method based on min-cut algorithm. Firstly,generate a point cloud graph. Then,choose the point which is going to be divided and set the threshold and the segmentation radius based on two points＇ Euclidean distance and color space distance. Secondly,find other points that in the range of designed radius and its＇ color and geometry information is below the threshold. Thirdly,use the region growing algorithm to find other points.Experimental results demonstrate that this method can efficiently solve the connectivity problems in 3d segmentation.
出处
《智能计算机与应用》
2017年第5期10-12,共3页
Intelligent Computer and Applications
关键词
最小割
彩色点云
分割
min- cut
colored point cloud
segmentation
