Object based classification using segmentaiton - QGis and Orfeo toolbox

المشرف العام

Administrator
طاقم الإدارة
I would like to apply an object oriented classification of Land Use and Land cover changes based on image segmentation of a Landsat 8. I am using Linux Mint 17.1 (64bit) and QGis2.8.1 and Orfeo toolbox.

I have already performed all 4 steps of the Exact Large-Scale Mean-Shift segmentation (LSMS) for an object based classification in Qgis using Orfeo toolbox. The segmentations are working fine, but I don't find a way to integrate the shapefile generated in the step 4 of the LSMS to a classification tool in QGis.

After the segments were created in step 4 LSMS, I have added an extra column (IDClass) to the table of contents and then the training samples (ROIs) were selected for further classification. I have identified the Land Use of some segments and I added it manually to this table.

Is that the usual methods to select training sample for object based classification in QGis?

After performing those steps, I would like to know how to apply the properly classification. What would be the next steps for an object based classification?

I think I should apply an TrainImagesClassifier and then classify the image. How could I do that using QGis and Orfeotoobox?

Thanks for any help.



أكثر...
 
أعلى