How to compensate for motion in 2 static point clouds?

I have 3 pointclouds from a single moving trajectory run of my sensor (attached). In that I can clearly spot a lamp post that appears slightly moved (as expected) and I wanted to see how can I compensate for it. I have just been able to plot them (and a bit of exprementation in that script), but I am not sure how to say segment the point clouds to focus on the lamp posts and then maybe manipulate them to compensate for it. Any help is appreciated.

The point clouds are from same run, just three different frames that came one after other.

Answer 1

[Rasmita]

Hi,

As per my understanding, you are using a script that reads the point cloud data from mentioned files and displays them. You are noticing a lamp post that is slightly shifted. So, you want to segment the point clouds and compensate for the shifted lamp posts.

Note that to segment the point clouds and focus on the lamp post, you can use a clustering algorithm such as the Euclidean clustering algorithm in MATLAB's point cloud processing toolbox. The ‘pcsegdist’ function segments a point cloud into clusters, with a minimum Euclidean distance of ‘minDistance’ between points from different clusters. It assigns an integer cluster label to each point in the point cloud and returns the labels of all points.

Once you have identified the points belonging to the lamp post, you can manipulate them to compensate for their movement. One way to do this is to apply a transformation to the point cloud using the ‘pctransform’ function. As an example, if you know the exact amount and direction of the movement, you can apply a translation or rotation to the point cloud to compensate for it.

For further information, please refer to the below documentation links:

https://www.mathworks.com/help/vision/ref/pcsegdist.html

https://www.mathworks.com/help/lidar/ug/estimate-transform-between-pointclouds.html

Hope this helps!

Regards,

Rasmita