Interpolating LiDAR: same steps but different results?

Question

I'm new to eCognition, and today I tried to use LiDAR data in the program.

First, I converted LiDAR to new image layers: 1) Elevation (first return) and 2) Number of Returns (all returns). I got some pixels with 0 value so I tried to interpolated these two new layers.

First I did 'Interpolating LiDAR Number of Returns', everything worked fine, I got a new image layer called 'Number of Returns Interpolated' (where formally 0 values are updated and contain now the mean of the mean values of its neighbors).

Then I started 'Interpolating LiDAR Elevation First' with exact the same steps:

First I carried out 'chessboard segmentation' and created a new level called '_temp Level' then I carried out 'assign class' so the Mean Elevation First > 0 will be assigned to 'Objects with value'. However, a warning showed up and said 'Domain is empty'. How could this happen? I mean I indeed followed the same step!

The rule set was made by my institution, if I just execute the whole rule set then there are no problem, but if I re-write rule set and execute each individual child process, the problem I described above happens. Any tips?

Answer 1

It's because when you execute from the start, eCognition will delete all levels.

You should insert a Delete Levelxxx before every child process. Before your ChessBoard Seg & classification step.

First I carried out 'chessboard segmentation' and created a new level called '_temp Level' then I carried out 'assign class' so the Mean Elevation First > 0 will be assigned to 'Objects with value'.

--> you have added a classification. And if there is no object to classify, eCognition will tell you the domain is empty. So if you re-write a rule, you need to delete the object levels and classification you create with the previous rule (the one you re-write). It is not needed if you execute the whole process.