Classify ground points lasts very long

[Damian]

Hello

I noticed, that ground point classification is taking now much more time than it used to. About 6 months ago classification was less time consuming than dense cloud generation. Later I noticed that both consume the same amount of time.

Now I have a project with dense cloud that contains about 3 900 000 colorized points, 8800 cameras  (42 MPix). It took 8h to generate depth maps and another 11 hours for dense cloud generation. Unfortunately, after the upgrade to version 1.5.3. the ground point classification after 10 hours shows me, that it needs 2 days 13 hours more to complete the task. I split the project into smaller parts (ca. 1 500 000 points), but there is no difference in time estimation.  Earlier it was not a problem to work with so big projects.

What could cause such situation? Before I upgraded to 1.5.3. I was not able to generate dense cloud (what has never happened before) due to "Image size mismatch error."

Looking forward to hearing from you.

[Alexey Pasumansky]

Hello Damian,

Which classification parameters you are using and how large is the covered area (in sq meters)?

Do you mean that the dense cloud has 4 millions points? Or this is a number for tie points? Then what is the number of points in the dense cloud?

[Damian]

Sorry for mistake, you are right: dense cloud has 3 200 000 000 points, tie points - 3 900 000

Parameters: 10 deg, max distance 0.3m, cell size 30m. Dense cloud points are located in the area of ~20km2. It is a road that is 35 km long and the buffer is 550 m width.



Update: it took 30 hours to perform a classification in a project with dense cloud 1 300 000 000 points, 10km2, parameters as above.

[Damian]

Could I get some help with this issue? I am working with another project: Dense cloud 3 400 000 000 points, 7 500 000 tie points, dense cloud clasification after 7 hours 11%, 1 day 18 hours left.