Accuracy Assessment in Erdas Imagine 2015

Question

I'm trying to achieve accuracy assessment in Erdas, but I've faced some difficulties due to my lack of knowledge about it.

I've finished an object based classification in eCognition, but whenever I try to obtain an accuracy assessment, it always give %100 accuracy with TTA mask, so I gave up.

now, I've decided to get that accuracy results from ERDAS because of high reliability to ERDAS among the users.

Workflow:

I've got classified shapefile and the raster file I've used to achieve classification. in ERDAS, it asks for an .img file which is a thematic raster file it says. so how can I generate an .img file from my existing .tif and .shp files? or is it even necessary to execute that process?

Answer 1

For whom looking for the answer for this question, please check the link below;

http://shreshai.blogspot.com.tr/2014/08/exporting-ecognition-classification.html

Besides; You can export the classification directly from eCognition as an .img file.

Workflow: Use the algorithm "export thematic raster files" > Define the path > Select export type: Classification > then Execute.

now you have given "thematic raster file" to process in Erdas Accuracy Assessment tool.

*Be careful about the classes you're exporting. Null classes, even if you've not assigned any sample inside, going to affect your accuracy assessment results.