Detecting significant change using the M3C2 plugin

[bdanhoff]

First and foremost I apologize if the has already been addressed. I am relatively new to this and searched around the forum but didn't find questions/answers that seemed to address my issue.

My goal is to hopefully use the M3C2 tool to look for changes to substrates after sand is added (simulating embedding). Before using M3C2 I used the cloud to cloud distance tool to get a general idea of the change that occurred, and as can been seen there are some regions where the substrate has not changed much (<2mm) while in other regions there are greater differences

0-90difference.JPG (142.3 KiB) Viewed 15903 times

My hope was to use M3C2 to determine where significant change occurred (i.e. the regions with 2mm difference might not be significant compared to others), but every time I've analyzed the two clouds (using default settings for normals & projection = 0.013m, and 0.5m for max depth) I either get 100% significant change or 0% significant change for the entire model depending on if I'm using the precision map setting, which I'm pretty sure I'm not using properly. From the wiki page it says that the precision map can be done using the scalar fields from the cloud. When I check the box to use precision information the only option for the dropdown for Sigma(x), Sigma(y), and Sigma(z), are "Scalar field", rather than the sX, sY, sZ that are shown in the sample on the wiki page. Does this mean that scalar fields are not available for my data? The use of precision mapping is (I think) secondary to my initial question about what is considered significant change in the models.

Any help with this would be greatly appreciated (and please let me know if additional information is needed).

Regards,
Brian

[daniel]

Indeed, Precision Maps are relying on the 3 scalar fields (SigmaX, Y and Z) that gives an information about the uncertainty associated to each point. These scalar fields can be generated by some photogrammetry software or custom algorithms (outside of CC). Apparently there's no such scalar fields associated to your data. Therefore you can't use this feature.

Anyway, the M3C2 plugin is able to compute a measure of confidence (the 'significant change' scalar field). You should take a look of how it is computed it may help you understand why it's always 1 or 0... But maybe it's not adapted to your present case?

[bdanhoff]

Thank you for the clarification on not being able to use Precision Maps. I have been using Agisoft Photoscan and will look into this. I will also read through the original publication again to see if there is anything that would indicate why I'm always getting 0 or 1.

Regards,
Brian

[JWallace]

Hi Brian,
How did you fare with the M3C2 and precision map calcs?
I am starting my final year Uni project and I will need to use these tools.
I will be using Metashape.
Kind regards,
James Wallace

[Setina]

hie Jwallace,
Did you manage to perform change detection using M3C2 plugin? I am starting my final year as well and l would like to use this plugin.