Hi at all,
I don't have time to work on this and I'm happy that hardy made such a detailed review. So @hardy: please go ahead and fix it. And now, I don't know if the extent property was fixed in the meanwhile.

Thank you both for keeping care of this.

-- Andy

I have tested the files from Andreas (see his comment #15) and they work quite reasonably, already.

Regarding the question in comment #16: Andreas' current implementation shows the coordinates (1,1) for the top-left most black pixel in this example image. This is identical to what Matlab shows. Also the displayed coordinate value for the bottom-right most black pixel is identical to what Matlab displays. I think there is no need to improve the 0/1 coordinate origin in any way.

Nevertheless I have found some minor issues with this code:

• When the mouse position is outside of the image position this code now displays "no image". Matlab displays a different dummy text in this situation: "Pixel Info (X,Y) Intensity".
• The displayed text string currently has a surplus closed bracket at the end (at least in the case of grayvalue images).
• The function impixelinfo does not return any handle (it should return a handle to a uipanel object). This could easily be fixed by returning the already existing h variable.
• The function impixelinfoval does also not return any handle (it should return a handle to a uicontrol object). This could easily be fixed by returning the already existing txt variable.
• The function impixelinfoval checks if it is called from inside impixelinfo and then generates a different text string (including the additional "Pixel Info" string). This seems overly complex to me. Could we not skip this check in impixelinfoval, and then add this "Pixel Info" string at the beginning of the displayed string inside of the calling impixelinfo.m? Or is there a good reason to do it this way?
• There are also a couple of "fix me" issues mentioned in the code itself:
  • The documentation text at the beginning of the file is missing in both files.
  • Indexed images do not yet work.
• I see the error message "error: get: H must be a graphics handle" when running the test_it.m script and using the mouse in figure 3 (with the 2 subplots).

I think most of the issues could easily be fixed. And I would like to see these two new functions included into the image package.

@Andreas: Do you know if the Octave issue with the "extent" property, that you mention in your code, has already been fixed in the meantime?

How should we proceed here?

• Is anyone currently still working on this?
• If not: Is anyone willing to work on this with the prospect of having this included in the image package eventually?
• If also not: Would anyone mind if I changed the code myself? (Only in case I find the necessary time before someone else does.)

@Avinoam: I forgot these patches... Or as far as I remember I lost track because I tired to hunt a bug with different text extent (which was a part of these patches)

I've created a repo on github: https://github.com/octave-de/octave-impixelinfo

If someone help to polish it send a short message with github user and I#ll give write access.

I didn't know that the axes could return the currentpoint in axes data coordinates (I used figures currentpoint), which would have saved me a lot of trouble as I had to change figure coordinates into axes coordinates  (which depends on many things ...).
Andy's implementation is much cleaner and simpler. Forget about mine.

@Andreas
It looks like you've got it pretty much in hand, so please go ahead. I was going to take a different approach, and am not nearly as far along as you are.

Thanks!

I've splited the function as discussed with carandraug, see attached tgz. the doit.m is just a helper to check for BW, RGB, grayscale images and so on.

(file #37757)

I started implementing impixelinfo last year. I'll attach the last version

(file #37751)

Yes, you're welcome to ask any questions.  I won't necessarily know the answers, but someone else may.

Thanks for this patch, Eric.  I have been wanting something like this!

Do you have any time to refresh the patch and implement impixelinfo and impixelinfoval?  It would be great to get this into Release 4.2.0, but that goes into feature freeze in a week or so.

This should be done optionally and not by default. Specifically, this should be the job of the missing functions impixelinfo and impixelinfoval (which would be part of the image package).