Structured versus Unstructured for Change Detection

[madihiberg]

Hey all!

The research group I am part of does work with data collected from both drones (photogrammetry) and terrestrial laser scanners. For my grad project I have been working with data collected by a Z+F 5010x scanner to do change detection analysis. All of our data editing is done in Autodesk ReCap pro, so to bring it into CloudCompare I export it to an E57 format. My question is surrounding the export option of 3d-only or complete within ReCap and what that means for doing change detection analysis?

The two options for exporting the data are “complete” includes per-scan info (no edits), and “3d-only” includes all edits (unstructured). Our issue is that we do not know what the difference between the 3d-only unstructured data and the complete structured data from the laser scanner actually means? I've been doing my change detection analysis with the 3d-only option as I assumed that means it would be exported with the edits I made in ReCap (removing noise, people, objects from the point cloud I don't want) and it seems to be working fine. Other members in the research project are concerned though with not knowing exactly what sort of information is being given up changing to the unstructured format because, again, we are not really sure what the difference is in this export option. I have attempted to upload our data after exporting in the "complete" option but the file sizes are too large and crash CloudCompare before they can be fully opened. So my main questions here are:

- What is the difference between these two export options from ReCap?
- Will change detection analysis work the same on either 3d-only or complete exports?
- Which format should be used for change detection analysis?

We are newer to using CloudCompare so I appreciate your kindness if we are asking very obvious questions, and thank you for your help!

[daniel]

So it's just about additional data that come with the 3D points.

With the "3d-only" mode, I guess you export only the 3D points. Which is indeed what is called "unstructured".

With the other mode, you probably get the grid structure of each scan and the position of the sensor relatively to the points. It's very useful to compute robust normals, but it can also be used to filter out some false detection cases when comparing point clouds.

The issue for you, is that this structure is properly supported/imported by CloudCompare from Faro, DP or PTX files... but not from E57 files!

So for now (supporting the scan structure from E57 files is in the TODO list), you can keep the "3d-only" mode. It probably makes smaller files. But whenever this is supported, you should switch to the "complete" mode.

[madihiberg]

Thank you so much for your reply and help!