Georeference Point Cloud in CloudCompare

    Intro

    Rock Mapper allows the collection of coloured 3D Point Clouds using the iPad Pro's built-in LiDAR sensor and camera array. These Point Clouds can be georeferenced right within the app. However, in some occasions you might want to georeference the point cloud later in the office. 

    CloudCompare is a free & open-source software package that allows the 3D registration of point clouds based on (at least) 3 known points. For example, you have scanned an underground rock wall or rock face that has 3 survey points marked up on it. You could then skip all georeferencing steps in Rock Mapper and instead perform the 3D registration later in CloudCompare. 


    ℹ️ Rock Mapper uses an internal coordinate reference system for point clouds. The position of the iPad Pro at the time when the 3D scan was initiated is set as origin with X/Y/Z = 0/0/0.

    ℹ️ If you are using CloudCompare on a regular basis and/or commercially, then we encourage you to support the open-source project with a donation.

    Workflow

    ℹ️ The workflow described below is for LiDAR Point Clouds, but also works for LiDAR textured meshes.

    1. Open Point Cloud

    Open CloudCompare and then choose File > Open




    Pick the [FaceID]_pointcloud.csv file from the uploaded Rock Face ZIP file that you would like to georeference. You'll then see the Open Ascii File window. CloudCompare automatically detects the correct columns for X/Y/Z and RGB (colour). 

    Just make sure to set Skip Lines to 1 so that the first line with column headers is excluded. Additionally, Ignore the HEX_COLOUR column. By default CloudCompare will import it as type label which can lead to issues (CloudCompare might freeze). Then press Apply.



    The point cloud will then be shown in CloudCompare's 3D Viewer. 




    2. Align point cloud based on known points

    Select the Point Cloud in the DB tree and then click the Align Tool icon in the top toolbar




    A new window appears that will now allow you to pick the points for georeferencing on the point cloud. 



    Pick the first point on the point cloud. It will be labelled as A0 and added to the table. Press X button to delete point in case it was incorrectly picked and then choose a new point. 



    Continue selecting points. Pick a minimum of 3 points. They will be shown as A0, A1 and A2 in 3D and in the table. 



    The real world coordinates (as picked up by survey) have been stored in a CSV file and are now used in CloudCompare.




    Cross_LW: Cross on left wall (A0)

    Bolt_LW = rock bolt on left wall (A1)

    Bolt_RW_low = lower rock bolt on right wall (A2)


    Click the pencil icon in the reference table to add the coordinates for the 3 points. 



    Now enter / copy & paste the coordinates for the first point. Please note that the order of these points has to match the order of the picked points. 



    Coordinates for point A0 (Cross_LW) entered


    Press OK and then repeat for the two other points (A1 and A2). Please note that when clicking the pencil icon the second time, the window with the X Y Z coordinates will show/remember previously entered values. Simply overwrite. 


    You should now see to original coordinates of the picked points in the upper table (A0, A1 and A2) and the entered reference point coordinates in the lower table (R0, R1 and R2). 




    Now press the align button and the point cloud will be aligned to match the reference point coordinates. Press green tick mark to close the alignment window.




    It will then show you the transformation matrix and an RMS error value. In this case the error is around 1cm, which is very good.



    3. Save / export Point Cloud

    You can now export the georeferenced/aligned  point cloud. Make sure to select it in the DB tree first, then choose File > Save in the top menu bar.



    ℹ️ Store as ASCII cloud in CSV format or as LAS file depending on the software you are going to use the point cloud in.




    Published