🚀 go-pugleaf

You can try to detect the colour pixels (if you look at the RGB values, 
colours are values where R,G,B is not close to equal). Once you've flagged 
these, you can use an inpainting algorithm to reconstruct the underlying 
gray values. I do have code for both, but not free. You can also google for 
"inpainting", and see what you get (this step is the hardest).

Nils

"Adem" <adem.meda@gmail.com> schreef in bericht 
news:488a9e8e@newsgroups.borland.com...
>I have access to jpg files of scanned documents.
>
> The documents themselves are public stuff (so, there's no copyright),
> but the jpg files contain watermarks.
>
> These documents are all typewriter documents, IOW black-white (or
> grayscale), and the watermark is color.
>
> Does anyone know of a way that can remove these watermarks. 

Nils Haeck wrote:

> You can try to detect the colour pixels (if you look at the RGB
> values, colours are values where R,G,B is not close to equal). Once
> you've flagged these, you can use an inpainting algorithm to
> reconstruct the underlying gray values. I do have code for both, but
> not free. You can also google for "inpainting", and see what you get
> (this step is the hardest).

Nils,

Thank you. 

I have found the solution I was after --albeit through some sort of
brute force..

Apparently other people have been annoyed by the same thing and they
have removed those silly watermarks. I have located those copies.

Thank you again,

Cheers,

Adem

That shouldnt be able to be done !

DH 

David HAROUCHE wrote:

> That shouldnt be able to be done !

I have just this kind of in the back of my mind at the time I wrote
that post: Someone would come up and imply that I was (or might be)
--in some way-- aiming to act or perform something unethical(ly).

I have seen such reactions in the past and I don't really blame them
--I have, at times, reacted likewise too.

But, life isn't always fair: I encountered a case where they
watermarked scanned images just because they could --not because they
had any right (copyright or otherwise).

Here's the longer story: 

A public office releases some documents in PDF form. These PDFs (that
contain purely scanned images of typewritten documenst) are not
available on the Net. On those PDFs, the public office has no copyright
claims whatsoever.

Someone, who has/had access to those PDF files, extracts the scanned
images and places them on the Net, but watermarks them for no reason
--but, makes it impossible to OCR the thing.

Now, would I, in your opinion, be doing anything unethical if I removed
their watermark?

[Luckily, as I said in the other thread, someone else released these
images without the watermark mess, saving me a lot of headaches.]

"Adem" <adem.meda@gmail.com> wrote
> A public office releases some documents in PDF form.  
> On those PDFs, the public office has no copyright
> claims whatsoever.
> Someone, who has/had access to those PDF files,
> extracts the scanned images and places them on the
> Net, but watermarks them for no reason --but, makes
> it impossible to OCR the thing.
> Now, would I, in your opinion, be doing anything unethical
> if I removed their watermark?

Perhaps, that "someone" would like and should be compensated
for their work for acquiring the documents, doing the conversion, 
and making the documents available on the net. --JohnH

I preffer to produce the documents by myself, doing file/import operation or 
pixels[x,y]

Regards

DH 

John Herbster wrote:

> Perhaps, that "someone" would like and should be compensated
> for their work for acquiring the documents, doing the conversion, and
> making the documents available on the net. --JohnH

Sure they are. They are getting plenty of hits whereby they make people
watch ads and what not.