python:建立KDTree并实现KNN搜索
方法1:【使用sklearn构建KDTree】
import sklearn
from sklearn.neighbors import KDTree
# build KDTree and KNN
KDTree = sklearn.neighbors.KDTree(pcd)
dist_to_knn, idx_of_knn = KDTree.query(pcd[idx], k=k)
注:pcd[idx]可以不是pcd中的点
方法2:【使用open3d构建KDTree】
import open3d as o3d
# o3d-knn
pcd = o3d.geometry.PointCloud()
pcd.points = o3d.utility.Vector3dVector(pcd)
pcd_tree = o3d.geometry.KDTreeFlann(pcd.points)
[k, idx_of_knn, dist_to_knn] = pcd_tree.search_knn_vector_3d(pcd[idx], k)
注:
由于open3d中search_knn_vector_3d函数的定义,输入的坐标只能是[3,1],所以查询点无法批量查询
search_knn_vector_3d(self, query, knn) Parameters query
(numpy.ndarray[numpy.float64[3, 1]]) – The input query point. knn
(int) – knn neighbors will be searched. Returns Tuple[int,
open3d.utility.IntVector, open3d.utility.DoubleVector]