写出python代码open3d导入ndimage模块、读取pcd点云数据、设置NDT算法参数、运行NDT算法、提取线状目标，可视化线状目标并保存结果

以下是Python代码示例：

import open3d as o3d
from scipy import ndimage

# 读取点云数据
pcd = o3d.io.read_point_cloud("point_cloud.pcd")

# 设置NDT算法参数
voxel_size = 0.05
max_iter = 50
max_distance = 0.1

# 运行NDT算法
reg_p2l = o3d.registration.registration_ndt(
    pcd, pcd, voxel_size, ndimage.uniform_filter, max_iter,
    [o3d.registration.TransformationEstimationPointToPlane()],
    o3d.registration.ICPConvergenceCriteria(max_iteration=1e-6, relative_fitness=1e-6, relative_rmse=1e-6),
    max_correspondence_distance=max_distance
)

# 提取线状目标
line_set = o3d.geometry.LineSet.create_from_point_cloud_correspondences(
    pcd, pcd, reg_p2l.correspondence_set
)

# 可视化线状目标并保存结果
o3d.visualization.draw_geometries([line_set])
o3d.io.write_line_set("line_set.obj", line_set)