masking and histogram functions?

[h0ughy]

This is a question to do with processing, can you stretch an image with the histogram function, and then apply a mask to the feature you want to keep, and then invert the mask protecting the feature you brought out and “deintensify” the background?

[dhalliday]

Boy...I hope so...because I do it ALL the time...

Its not really accurate,because the mask favors the stronger signal,but if done in steps,I love it.
I suspect that technically its not up to "PI snuff"....will wait to hear from others...
Isn't this a bit of what the masked stretch utility is doing...??

Dave

[h0ughy]

Boy...I hope so...because I do it ALL the time...

Its not really accurate,because the mask favors the stronger signal,but if done in steps,I love it.
I suspect that technically its not up to "PI snuff"....will wait to hear from others...
Isn't this a bit of what the masked stretch utility is doing...??

Dave

so you can do it?  Can you spell out in detail what you do?  a few screen shots?

[Niall Saunders]

can you stretch an image with the histogram function, and then apply a mask to the feature you want to keep, and then invert the mask protecting the feature you brought out and “deintensify” the background?

Yes, in theory you can, but the difficulty is in finding a Histo stretch that will correctly 'deintensify' the previously stretched image. In other words, you need to find the 'inverse' of your first stretch.

However, if you stretch the image as planned, to get the area you are interested in looking the way you want it to, and then make a duplicate (clone) of this modified image, and then make a mask from this cloned image - revealing only those areas that you want to change.

Now, go back to the first, modified, image and 'undo' the histo transform - don't worry, the Histo process still 'knows' what your setup parameters were. Now you apply the 'mask' to the original image, and then re-apply the Histo stretch 'through' the mask.

Is this what you are after?

Cheers,

[Juan Conejero]

Hi,

In other words, you need to find the 'inverse' of your first stretch.

This is actually very easy. If you have applied a HistogramTransformation (HT) with a midtones balance m, then the inverse transformation is 1 - m. For example, applying two successive histogram stretches with m=0.1 and m=0.9 leaves the image —neglecting roundoff errors— in its initial state.

Of course, the above does not apply if the image has been clipped at one or both ends of the histogram. Clipped pixels are unrecoverable. But as long as you only apply midtones transformations, the inverse transformation of m is 1-m.

[Niall Saunders]

the inverse transformation is 1 - m

I had never given that any consideration Juan - I have learned something new again (which, of course, now means I have to 'unlearn' something 'old', as my neural net has reached full capacity - and if I don't watch out, I will exceed my full-well capacity and leave horrible streaks all over my mind )

Cheers,

[jmtanous]

Juan,

Thanks for the tip!

Another technique to abuse my images!

Cheers,

José