Nearest Neighbour Classification in eCognition

Question

I have a few papers and youtube videos explaining how to conduct the NN classification in eCognition but it isnt giving me any output.

I think the confusion is arising when trying to define the statistics in eCognition Developer.

Could someone like a step by step guide for dummies or explain it to me in very simple terms?

Answer 1

I think I had the same issue. First check that all your classes including unclassified class, are selected in your Class filter (see image). I hope that will solve your problem. I have to say as great as eCognition is, it's interface does tack some time to know

for a great step by step guide look at this page

This is one of the best GIS/RS blogs.

one tip when doing a NN classification, have a look at the use of the feature space optimization tool. It will give you mush better results if you do that before the NN classification.

Answer 2

You should give some more background; What I do, once there's a loaded image:

1. In the process tree window: right click: append new > multiresolution segmentation, parameters: Scale = 100;
2. Classification > class hierarchy; add your classes
3. Classification > Nearest Neighbor > Edit NN Feature space; add your values, such as layers>mean or texture by double clicking
4. Apply standard to NN classes
5. Training: Classification>Samples>select samples -for each class;
6. In Process tree tab, right click: append new>classification and execute