Can I use eCognition to check which bands are more helpful to separate individual classes and to get output result? If so, how can I do this?

  • What is the connection between your ArcGIS tags and your question? – PolyGeo Sep 4 '13 at 9:47
  • how matrix error on ecognition. how can help me. mail: es.rad1204@yahoo.com – user35601 Aug 9 '14 at 10:49

U can do this task in eCognition. The process steps are

  1. Do segmentation; preferably multiresolution (of scale parameter 5) or chessboard segmentation ( of scale parameter 1; this will useful to understand the pixel values)
  2. Now in the Feature View, you can see the Object features >> Layer values >> Mean >> in which your uploaded image layers.
  3. Double click on each layer. On the down right corner click the check box,then the values can be seen.
  4. You can select which spectral band is useful to detect each classes. You may split the windows by window >> split horizontally >> side by side view. Now update the values of each image layers on different windows. You can compare the results of each band.
| improve this answer | |
  • This helped me a lot. Thank you very much, Sandhya! – Bandrush Barda Sep 4 '13 at 11:43
  • eCognition is one of the best classification software. It is useful to complete complex task. Agree? – Sandhya Sep 4 '13 at 12:37
  • Totally agree with it. It seems quite powerful software for classification. Recently, I'm very interested in eCognition. Actually I'm one of the new users of it. – Bandrush Barda Sep 4 '13 at 16:39
  • By the way, how can I get class separability report between the classes, such as M-statistic. I suppose the separability report would be very similar with cross tablulation in accuracy assessment. – Bandrush Barda Sep 4 '13 at 16:45
  • sorry, similar with error matrix, not cross tabulation. – Bandrush Barda Sep 4 '13 at 17:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.