I am studying on concrete technology, I have jpg image of concrete which is taken by digital camera .I am working with arcgis 10.1 version. I need to convert it into greyscale and get attribute table for further analyasis and classification.How can I get that ? enter image description here

  • Python's scikit-image library is really a slick way to perform digital image processing such as this. scikit-image.org – Aaron Jul 3 '17 at 3:16

This is a workflow with ArcGIS 10.3 + Spatial Analyst, so it can be different in your case.

  1. Open ArcCatalog and read each band of the F4Mmj.jpg(this name came with the attached image when I downloaded it) into ArcGIS (without spatial reference)... actually just drag & drop Band_1, Band_2, Band_3 onto the map window.
  2. Open Raster Calculator and run this expression ("F4Mmj.jpg - Band_3" + "F4Mmj.jpg - Band_2" + "F4Mmj.jpg - Band_1") / 3.
  3. Conversion Tools | From Raster | Raster to ASCII to export the calculated raster layer out to ASCII text file.

My preference, however, is to use ImageJ for image conversion, as it does not require Spatial Analyst license.

  1. Open the jpg file by ImageJ, and select Image | Type | 8-bit ... this will change the RGB to 8-bit greyscale. (This process also uses the same equation as above, (R+G+B)/3).
  2. Save it as new image file and give it to ArcGIS.
  • Thanks @OnkarKatgaonkar ; As to my 3rd point, I assumed that you needed pixel values. Please let me know if it is not the case. Hopefully this can answer to your previous question: How to access attribute table of jpg image? , too. – Kazuhito Jul 2 '17 at 7:48
  • I get greyscale image successfully , but I have some query about pixel value classification, because concrete have cement,fine and coarse aggregate. How can I classify this material through greyscale? I need an unique identity pixel value for each material classification – Onkar Katgaonkar Jul 2 '17 at 20:39
  • @OnkarKatgaonkar Interesting, but I honestly do not know methods through greyscale. Common image classification approach would require RGB spectral signature. You may want to post new question. – Kazuhito Jul 2 '17 at 22:25

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.