🔶Point Cloud Ordering

Function Description

Using camera intrinsic parameters, converts input, usually unordered camera coordinate system point clouds into ordered point clouds.

The converted point cloud structurally forms a two-dimensional grid, where three-dimensional points from the original point cloud are mapped to corresponding positions in this two-dimensional grid.

If some positions in the grid do not have corresponding three-dimensional points in the original point cloud, these positions will be marked as invalid values (NaN).

Usage Scenarios

  • Alignment with 2D Images: Associate 3D point cloud data with 2D images taken by the same camera at the same viewing angle at the pixel level, for example, finding the corresponding depth value or 3D coordinates for each pixel in the image, or assigning image colors to point clouds.

  • Generate Depth Maps: The Z coordinates (depth information) in ordered point clouds can be directly extracted or converted to generate a depth map aligned with the reference image after simple transformation.

  • Apply 2D Image Processing Algorithms: Some filtering, feature extraction, and other algorithms designed for 2D images can be applied to the converted data structure after point cloud ordering (for example, applying 2D mean filtering to depth values of ordered point clouds).

Input Output

Input

Image: Reference image.

Camera Coordinate System Point Cloud: A list of one or more point clouds.

Camera Intrinsic Parameters: Camera intrinsic matrix.

Output

Ordered Point Cloud: Point cloud after ordering processing. If the input is a point cloud list, the output will also be a list of ordered point clouds corresponding to each input point cloud.

Parameter Description

This operator has two versions:

  • Point Cloud Ordering: Processes point clouds without normal information.

  • Point Cloud Ordering (with normals): Processes point clouds with normal information.

Both have identical core functions and parameters, differing only in the type of point cloud data they process.

Merge Input Point Clouds

Parameter Description

Used to control when inputting a point cloud list, whether to first merge all point clouds in the list into a single, large point cloud, then perform ordering processing.

Parameter Adjustment

  • Disabled (default): If the input point cloud is a list containing multiple point clouds, the operator will independently perform ordering processing on each point cloud. The output ordered point cloud will also be a list containing the corresponding number of ordered point clouds. Suitable for situations where multiple independent point clouds need to be ordered separately.

  • Enabled: If the input point cloud is a point cloud list, the operator will first combine all point cloud data in this list into a single point cloud. Then perform ordering operation on this merged point cloud. The output result will be a list containing only one ordered point cloud. Suitable when multiple point cloud fragments of one scene need to be viewed as a whole to generate a unified ordered view, this option can be enabled.

Align Image Size

Parameter Description

This parameter controls whether the two-dimensional dimensions (width and height) of the output ordered point cloud strictly match the input image dimensions.

Parameter Adjustment

  • Disabled (default): The output ordered point cloud dimensions will be determined based on the coverage range of actual point cloud data projected to the image plane through camera intrinsic parameters. This means the output point cloud width and height may not exactly match the original image dimensions, but will ensure all effectively projectable points are contained in a compact two-dimensional grid as possible.

  • Enabled: The output ordered point cloud width and height will strictly match the input image dimensions. In this case, if some three-dimensional points project outside the image range, these points will not appear in the output ordered point cloud. Suitable for scenarios where the generated ordered point cloud (such as depth maps) needs to correspond precisely pixel-by-pixel with the original image, this option can be selected.