1

I have been tying to implement this paper on an SVM classifier wherein I need to extract training samples from an image and run an SVM classifier on it. The paper says the following:

IDRISI Andes was used for the preliminary processing such as identification of training samples. These data were then exported to MATLAB (version 7) for SVM classification.

However, I cannot figure out how to export the training sample data in a format readable by MATLAB (or even python, for that matter). The help page I found doesn't seem very useful.

I found answers here that tell me eCognition has a utility to do this, and so does grass GIS. Is it recommended that I use one of these softwares, or is there any way to do it in IDRISI?

(I also found this question which does not have any answers.)

1 Answer 1

1

I guess they used Idrisi only to define geographical aspect of the training samples. In this case they could have exported only the shapefile that can be read by MatLab as well as Python (GDAL) along with its attributes. On the other hand class signatures can be exported as a plain text and therefore you will need to write some kind of a parser for the software to understand the numbers written in the file.

edit: QGIS is free and really easy to use. If you don't know how to create a shapefile there are many starter tutorials, just google it. But you could create it in any GIS. Idris exports shapefile using export > vector > software specific format > ESRI shapefile or something similar.

1
  • Oh, in that case, is there a simpler software I can use just to obtain the geographical aspects of the samples? I am finding it very troublesome even to load images into IDRISI.
    – loudmummer
    Commented Jun 8, 2018 at 8:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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