RCStructuredScan: are some points definitively lost by Recap?

[lamorlette]

Sum-up:
It seems that Recap lose points because of some spherical remapping, and these points are definitively lost.

Details:

Recap SDK provides two ways to read a scan which give different points:

• The unstructured reading via IRCPointIterator gives the same points as one can see in Recap application. But these points are not the one of the original file: they are decimated and filtered.
• The structured reading via RCStructuredScan gives the same points as the ones of the original file, but some (around 10% in my example) are missing: one can see slanted stripes of invalid points when generating a panoramic image. It seems that these missing points are spherical remapping artefacts.

One can see this in the following images:

• Original E57 file (as seen in Cloud Compare):
  
• 3D view in Recap:
  
• E57 export from Recap (same as original file, but with missing points):
  
• Unstructured reading via IRCPointIterator (same as 3D view in Recap):
  
• Structured reading via RCStructuredScan (same as E57 export from Recap):
  

Questions:

1. Is my understanding about decimation and filtering in Recap correct?
2. Is my understanding about spherical remapping artefacts for E57 export from Recap and structured reading via Recap SDK correct?
3. Is there a way to recover the points missing in structured reading, or are they definitively lost?

[yan.fu]

Hi @lamorlette ,

Thanks for your detailed information here. 

The original points that are not mapped to the structured scan (.rcc file) are saved into .diff file (which can be found from the project support folder). The points got from unstructured point iterator might be decimated or not, depending on the decimation grid value set during the scan importing. Because the minimum decimation grid size is 1 mm, if the original point cloud is very dense, then the point cloud might be decimated. Otherwise, all the points should be saved into the .rcs files. 

Best,

Yan

[lamorlette]

@yan.fu 

---

The original points that are not mapped to the structured scan (.rcc file) are saved into .diff file (which can be found from the project support folder).

---

Is there a way to get those original un-mapped points with the Recap SDK?

---

The points got from unstructured point iterator might be decimated or not, depending on the decimation grid value set during the scan importing.

---

When creating the RCP from the E57, I chose the “advanced - decimation disabled” option. Still, it seems to me that the points one can see in Recap are decimated.

---

Because the minimum decimation grid size is 1 mm, if the original point cloud is very dense, then the point cloud might be decimated. Otherwise, all the points should be saved into the .rcs files. 

---

Even with the “decimation disabled” option, the points displayed in Recap and returned by IRCPointIterator seem not only decimated, but also filtered (see the screenshots). Please, what do you think about this?

[yan.fu]

Did you set the noise filtering option to be minimum?

Yan

[lamorlette]

To create the RCP project I gave as example (see screenshots above) I used Recap application with the “scan settings” - “advanced” - “decimation grid disabled” option:

I can't see any noise filtering option.