How to identify image type is halftone or line art

[Karthik G Kumar]

Hi,

      How to identify image type is halftone or lineArt. Please help me.

Thanks in advance .

[Test Screen Name]

Zoom in to 100% or maybe 400% to look for a halftone pattern. Use your knowledge. I do t think any software will tell you. Be aware that in very old illustrations, halftoning was handmade, so it would be both. 

[Stephen Marsh]

Can we presume that both files are bitmap mode, 1-bit files?

[Karthik G Kumar]

Hi,

        How to identify image type is halftone or lineArt in photoshop using extend java scripts. Please help me.

Thanks in advance 

[c.pfaffenbichler]

Please post examples. 

Technically »lineart« should be bitmap/1bit images but I am not sure if you are referring to that simple criterium. 

[Stephen Marsh]

Duplcate post:

https://community.adobe.com/t5/photoshop-ecosystem-discussions/how-to-find-the-images-type-is-halfto...

[Karthik G Kumar]

Hi,

How to find  the Images type is Halftone or LineArt in Photoshop using extend java script. Kindly help me .

Thnks in advance.

[Bernd Alheit]

Try the forum for Photoshop.

It is not possible with a script in Adobe Acrobat.

[JR Boulay]

[MOVED TO THE PHOTOSHOP DISCUSSIONS]

[Stephen Marsh]

Dupe post, perhaps a moderator could merge?

https://community.adobe.com/t5/photoshop-ecosystem-discussions/how-to-identify-image-type-is-halfton...

[Stephen Marsh]

I know of no way to detect this with an action or script.

This would likely take machine vision/learning, AI etc.

In the frequency domain, the repeating patterns formed by a regular halftone screen do create a different image than a lineart image.

Example:

So potentially, using an FFT transform plugin, post-processing and histogram evaluation, I guess it might be possible for this to be scripted.

EDIT:

Perhaps a simple evaluation of the histogram of both images converted to 8bpc may be enough, if you can find consistent weighting of the STD-DEV, MEDIAN or MEAN values of one type of image vs. another. I would wish to test using a wide variety of images though, both lineart and halftone... So the FFT suggestion may be taking this too far, it just came to mind as the repeating pattern is visually obvious, but perhaps not needed or helpful for histogram evaluation.