我想做深度学习跑点云配准，有没有相关代码，我是window11

您可以使用Python编程语言和深度学习框架TensorFlow或PyTorch来实现点云配准。以下是一些常用的点云配准算法和相关代码：

1. Iterative Closest Point (ICP)：ICP是一种基于迭代的点云配准算法，它通过迭代寻找最小化点云间距离平方和的变换矩阵。您可以使用开源库Open3D来实现ICP配准，它支持Windows操作系统。以下是一个ICP配准的示例代码：

import open3d as o3d

# Load point clouds
source = o3d.io.read_point_cloud("source.pcd")
target = o3d.io.read_point_cloud("target.pcd")

# Perform ICP registration
icp_result = o3d.registration.registration_icp(
    source, target, 0.02, np.eye(4),
    o3d.registration.TransformationEstimationPointToPoint(),
    o3d.registration.ICPConvergenceCriteria(max_iteration=200))

# Print transformation matrix
print(icp_result.transformation)

2. PointNetLK：PointNetLK是一种基于神经网络的点云配准算法，它使用局部特征描述符来匹配点云。您可以使用开源库PointNetLK来实现PointNetLK配准，它基于PyTorch实现并支持Windows操作系统。以下是一个PointNetLK配准的示例代码：

import torch
import open3d as o3d
from pointnetlk import PointNetLK

# Load point clouds
source = o3d.io.read_point_cloud("source.pcd")
target = o3d.io.read_point_cloud("target.pcd")

# Convert point clouds to tensors
source_tensor = torch.tensor(source.points).unsqueeze(0)
target_tensor = torch.tensor(target.points).unsqueeze(0)

# Create PointNetLK model
model = PointNetLK(feature_transform=True)

# Perform PointNetLK registration
model.eval()
with torch.no_grad():
    T = model(source_tensor, target_tensor)

# Convert transformation matrix to Open3D format
T = T.cpu().numpy()
T = np.vstack((T, np.array([0, 0, 0, 1])))
T = o3d.geometry.create_mesh_coordinate_frame(size=0.1).get_rotation_matrix_from_xyz((np.pi/2, 0, 0)) @ T

# Apply transformation to source point cloud
source.transform(T)

# Visualize result
o3d.visualization.draw_geometries([source, target])

以上是两种常用的点云配准算法和相关代码，您可以根据自己的需求选择适合的算法和框架进行实现。