| View previous topic :: View next topic |
| Author |
Message |
vulpes
Joined: 26 Aug 2004 Posts: 3
|
Posted: Thu Aug 26, 2004 3:37 am Post subject: Pixel --> Text? |
|
|
I know this might be asking too much,
but would it possible to convert difference in pixels into text?
Like read all the black pixels in a rectangular region and
translate that into text.
I'm figuring its gonna require a whole bunch of code ,
but just thought I should ask |
|
| Back to top |
|
 |
Chris Site Admin
Joined: 02 Mar 2004 Posts: 10464
|
Posted: Thu Aug 26, 2004 4:14 am Post subject: |
|
|
It'd be a great feature, but a lot of work. Perhaps there is some open source OCR software out there somewhere that could someday be integrated with AutoHotkey.
By the way, in addition to the normal OCR software, there is one utility I know of that actually "reads" text shown in any window by comparing it to the fonts you have installed. It's called Kleptomania (though it's not free). |
|
| Back to top |
|
 |
Guest
|
Posted: Thu Aug 26, 2004 8:26 pm Post subject: |
|
|
| Quote: | | Perhaps there is some open source OCR software out there |
Yep, Clara OCR ...
 |
|
| Back to top |
|
 |
Chris Site Admin
Joined: 02 Mar 2004 Posts: 10464
|
Posted: Thu Aug 26, 2004 8:58 pm Post subject: |
|
|
| Sounds nifty. Perhaps it has a command line interface that would make it easy to integrate into a script? Even if it doesn't, I guess you could automate its GUI. |
|
| Back to top |
|
 |
Jabberwock Guest
|
Posted: Sat Aug 28, 2004 12:06 pm Post subject: |
|
|
I think OCR would be a bit of overkill. 90% of the apps and desktops use about three fonts, so it would be enough to compare the relevant bitmap patterns. But it still would be lots of work.
I tend to agree with the Kleptomania view, almost all text displayed in applications (even hidden passwords!) might be captured in some way from the system itself. The obvious exception would be text written in a word processor, etc. |
|
| Back to top |
|
 |
Guest
|
Posted: Sun Aug 29, 2004 9:19 am Post subject: |
|
|
Program OCR is used for recognition Cyrillic letters. OCR is based on the backpropagation neural networks. OCR is developed as addition to masters thesis "Training neural networks for recognition Cyrillic letters".
[Download]
 |
|
| Back to top |
|
 |
|