Pipeline for using laser scan data as MTPs to improve accuracy


I have a site where roadway elevation is very important to capture. My experience with Pix4D is that it doesn’t do so well with elevation. Oftentimes, errors will build towards the edges of a project. This is very evident if you compare the drone point cloud to a laser scan point cloud, where the laser scan is known to have good elevation capture.

I want to use laser scan data as MTP’s to hopefully improve the accuracy of the drone point cloud and properly capture the elevation profile of the roadway that I’m interested in. I have tried doing this before but ran into some issues with the coordinate systems. When I obtain my laser scan point cloud, it is in an arbitrary coordinate system in feet, whereas Pix4D seems to like geographically-based coordinate systems. The last time I tried to do this using an arbitrary coordinate system, the project processed without using the GPS data as an initial guess for the camera positions. I would like to use both the GPS data and the laser scan data as MTP’s. Is there a pipeline for achieving this?

Thank you.

Hello EvanH,
You are correct in that the edges of the project will not reconstruct as well as the center of the project. We always recommend performing at least one flight line past the edge of the project area to ensure accurate reconstruction. Considering you are mapping a roadway that would be inherently linear you have a significant amount of edges to contend with. You will want to fly well past the edge of the roadway to ensure your accuracies are as best as possible.

Mapper does support arbitrary coordinate systems. You can read more about it in the link below.

You can also create a site calibration file if you are working in this file for multiple projects. You can read more in the link below.

Lastly, you cannot bring in the laser scanned point cloud into Mapper. Unfortunately it does not support this. You might be able to bring the laser scanned point cloud into a 3rd party software and create a point. But this isn’t possible with Mapper.