[SOLVED] How to get a text from an *imagen*?

Camello · 05-24-2021, 10:17 AM

Hello,

I need to convert an *image*, not a document, to a text so that it can be later downloaded in the GUI of an application.

I have tried uselessly for days now and as I has been oriented by others i tried using OCR and by the terminal too, and on websites ... nothing works.

I have consulted for days on Google, I have written to some places, I find nothing, apparently no one knows about this matter.

And such a tool exists, example, on this is this site below, there are already some images converted into a text, you just have to click on any of the images and automatically a text will be open in your browser then you can copy and ...

http://gorgonian.weebly.com/scid-vs-pc.html.

The image I want to get converting into a text we have to download it first, from here, the URL:

https://www.dropbox.com/sh/86q1rckd8...cid_pieces.zip

If someone here is aware of what a tool in question are worth for such a procedure ... such aid would be very much appreciated.

Thank you and a greeting.

business_kid · 05-24-2021, 10:32 AM

If you are converting from image to text, you will have to use one of two programs

Tesseract - OSS solution. Use 5.0 or higher, get your image in high res (≥600dpi) or expect poor results. Do a hand edit after.
Abbyy - proprietary software (= you pay) but it offers the best-in-class performance. Do a hand edit after.

You'll probably have to compile tesseract and it's dependencies, unless your distro does a package.

I used tesseract on a 1960s play strangely laid out on a manual typewriter with a dead ribbon but it came through well enough. I lost over a day mainly reformatting & correcting. There were also copious hand edits in fountain pen which didn't come through the OCR. You wouldn't have that hassle with other work.

Camello · 05-24-2021, 11:02 AM

"to compile tesseract and it's dependencies ...". No, i need something with less complications. Any other idea?. Thanks for your answer to my post.

EDIT: And as i said before i already tried OCR and it doesn't work because what i need it is to convert into a text not a document-image but a real image, exemple:

http://gorgonian.weebly.com/scid-vs-pc.html -- click on the image and you will see the text!.

teckk · 05-24-2021, 11:38 AM

Quote:

I need to convert an *image*, not a document, to a text

You can't. You are talking about 2 different things.

Look a that web pages source code.

First piece, top left
The text file that the hyper link points to
http://gorgonian.weebly.com/uploads/...habsburgos.txt

The image file for that hyperlink
http://gorgonian.weebly.com/uploads/.../656171627.png

Snippets from the page source.

Code:

<title>SCID vs PC - Gorgonian's Chess Site</title><meta property="og:site_name" content="Gorgonian's Chess Site" />
<meta property="og:title" content="SCID vs PC" />
<meta property="og:description" content="SCID vs. PC Piece Graphics " />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/656171627.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/954711183.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/2062471_orig.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/720022966.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/395224216.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/389885632.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/458283047.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/494460524.png" />
<meta property="og:image" content="http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/8194729_orig.png" />
...

<div><div class="wsite-image wsite-image-border-thin " style="padding-top:10px;padding-bottom:10px;margin-left:0;margin-right:0;text-align:center">
<a href='/uploads/1/7/2/2/17221082/habsburgos.txt'>
<img src="/uploads/1/7/2/2/17221082/656171627.png" alt="Picture" style="width:100%;max-width:442px" />
</a>
<div style="display:block;font-size:90%">HabsburgOS</div>
</div></div>

<div><div class="wsite-image wsite-image-border-thin " style="padding-top:10px;padding-bottom:10px;margin-left:0;margin-right:0;text-align:center">
<a href='/uploads/1/7/2/2/17221082/fritz.txt'>
<img src="/uploads/1/7/2/2/17221082/954711183.png" alt="Picture" style="width:100%;max-width:443px" />
</a>
<div style="display:block;font-size:90%">Fritz</div>
</div></div>

<div><div class="wsite-image wsite-image-border-thin " style="padding-top:10px;padding-bottom:10px;margin-left:0;margin-right:0;text-align:center">
<a href='/uploads/1/7/2/2/17221082/fritzsws.txt'>
<img src="/uploads/1/7/2/2/17221082/2062471_orig.png" alt="Picture" style="width:100%;max-width:448px" />
</a>
<div style="display:block;font-size:90%">FritzSWS</div>
</div></div>

You'll need to explain what you are trying to accomplish. Is there text on that image file?

Camello · 05-24-2021, 11:54 AM

I only give the previous link as a reference to indicate that it is possible to convert the images of the different chess pieces into a text as in that site they are offering some models but unfortunately they do not have what I am looking for: Leipzig

https://i.postimg.cc/qhVNvypL/leipzig-55.png

https://www.dropbox.com/sh/86q1rckd8...CID_Pieces.zip

More information:

i already got two styles of pieces from the website in question in my system but because they do not have 'LEIPZIG' i have been for days looking how to make the convertion by myself but till now i found not how to do it.

(the text it is included here, you don't need to do anything)

https://i.postimg.cc/4KbDtrrj/Captura-de-pantalla.png

shruggy · 05-24-2021, 01:08 PM

The problem is not as generic as the title of your thread suggests. The software that site uses is Scid vs. PC. Try asking on their mailing list.

Scid vs. PC is a fork of SCID. Your can try their mailing list as well.

teckk · 05-24-2021, 01:18 PM

https://sourceforge.net/p/scid/code/ci/master/tree/
https://sourceforge.net/p/scid/code/...pieces/Leipzig
https://sourceforge.net/p/scid/code/...ter/tree/help/

computersavvy · 05-24-2021, 01:18 PM

Your dropbox post is not accessible. Maybe you need it to be world readable. I can't even look at the file you reference. Or you could link the original source for that file so we could download it for ourselves.

Now as to the real problem. You say you want to convert the image to text. Why? Just to get the text that is part of the image? Or to convert the entire image to text the way images were sent over the internet in years past?

One other possibility might be to convert the image to a pdf then use a tool similar to OCR to grab the text from the pdf. I believe there are several tools that can do that although I am not sure they can work on embedded images.

teckk · 05-24-2021, 01:21 PM

It looks like the game is making some sort of hash for the images.

Look at:

Code:

curl http://gorgonian.weebly.com/uploads/1/7/2/2/17221082/habsburgos.txt

The OP will need to read the docs and source.

Camello · 05-24-2021, 01:58 PM

About the URL, for the porpouse it is the same that teckk link is giving here: https://sourceforge.net/p/scid/code/...pieces/Leipzig (it is the second link)

Another thing, without the convertion of this image, 'LEIPZIG', into a text the GUI (scidvspc) can't open the images/pieces.

EDIT: I don't know but i having problems to post the links here at this forum.

*It looks to me that if you do not download scidvspc and try to do what i'm looking for (to get the LEIPZIG style of pieces) you are not going to understand what i'm trying to explain here because my limitation in the computer field and in the english too. Thanks.

boughtonp · 05-24-2021, 04:18 PM

Quote:

Originally Posted by Camello

It looks to me that if you do not download scidvspc and try to do what i'm looking for (to get the LEIPZIG style of pieces) you are not going to understand what i'm trying to explain here because my limitation in the computer field and in the english too.

LQ isn't the support desk for scidvspc - if you're trying to use a feature of scidvspc you should be asking the developers/community of that software first - there are several options for doing so here: https://sourceforge.net/projects/scidvspc/support

Also, I'm not convinced the lack of understanding here is solely a language one, but since you're using MX Linux you could check if one of their other language forums is in your native tongue and maybe that will make it easier to explain what you're after.

Camello · 05-24-2021, 04:39 PM

As i said at the begining i have being trying many things, i post at MXlinux, etc, etc, ect ... Nodody know how to do it; and at sourceforge:

https://sourceforge.net/p/scidvspc/d...ad/0254cf4f8a/

Any way thank you.

business_kid · 05-25-2021, 08:29 AM

Post deleted.

Camello · 05-25-2021, 10:31 AM

Just in case. since I opened this thread ... i did't delete any post, any suggestion, advice, will be heard.

Camello · 05-27-2021, 10:11 AM

Finally, on sourceforge, someone kindly responded to the request:

https://sourceforge.net/projects/sci...2.tgz/download

Regards