Rasterization and general question

[ilovecoffee1125]

I use a program called SCENE that is associated with the terrestrial LIDAR scanner. I have about 80 comparisons dealing with bank erosion and deposition. Once I have processed the images in SCENE I then use CloudCompare to run the M3C2 algorithm that is needed to see changes. After running the algorithm, I then use the rasterization tool to export the associated active layer and then the associated heights of the scans. Of all the comparisons, 3 are showing strange height values. If the GPS data for one of the two comparison images is missing could this impact the results of the heights being exported, however, this has not impacted other scan height values. (As a few comparisons also have missing GPS data and their associated height values are correct).

Is Cloud able to alter height information or does it transform it in some way if all the required information it needs is unavailable?

In general, what values is Cloud using to get the heights that I export? (Altimeter, GPS?)

Also, would this missing information impact the algorithm?

Thank you,
Carly

[daniel]

Well, CloudCompare doesn't (shouldn't) alter the coordinates of the point clouds. In some specific cases, CC can ask you to shift the coordinates at loading time (if they are too big). But they are restored automatically at saving time (if the output format supports it). And in your case the problem happens visibly at loading time.

The only solution would be therefore that the coordinates are shifted at loading time. If your input files are FARO files, we trust the FARO SDK for reading the points out of the file. Maybe the way they handle/save the GPS information is the key?

Or do you use another file format to exchange data between Scene and CC?

[ilovecoffee1125]

Daniel,

I use pts files. The issue I am running into is that all of my other comparisons maintain proper heights throughout the entire process listed below. I use the same method for each comparison.

Do you think that this issue with the incorrect heights, or perhaps missing GPS information from the scanner, will impact the results from the M3C2 algorithm?

-Carly

[daniel]

PTS file should contain the sensor position, but I don't know if it's corrected or not with the GPS coordinates (everything happens on SCENE's side).

But of course, if the scans are shifted then it will definitely have an impact on the M3C2 computation... Do you have fixed features in your scan maybe? Can you use them to estimate the shift or even register the scans (with point-pair based alignment for instance)?

By the way, what is the accuracy of your GPS? And what is the approximate shift you observe?

[ilovecoffee1125]

There are targets placed in both scans that are stationary and are always in the same location. I use these to estimate the error. Most errors are around .004m. I believe the GPS is rather accurate on the scanner we use. It is fairly new. Like I mentioned before, I have other scans that are missing all GPS information and using a reference scan alone suffices for providing accurate heights.

[daniel]

Yes, but in the case where you observe strange height values, do the targets have the same very low error? I guess even the targets shouldn't be at the right altitude, do they? (I mean right at loading time).

Or did I miss something?

[ilovecoffee1125]

Daniel,

Yes, the scans with strange height values have relatively low errors.

Carly

[daniel]

Then I'm lost here ;)

There must be something wrong somewhere but it's hard to say what without the whole picture. At least I don't see how CloudCompare itself could explain this issue.