18 replies to this topic

[Xander]

100% pure AHK for recognizing the characters $.,0123456789

Download: ODR.ahk [ SKAN's Dropbox ]

Note that spaces are put in place of whatever character that doesn't map properly. To see the "signature" of unrecognized characters, uncomment the println(signature) line and include the following:

Print(str) { 
   static create = true, output 
   if create { 
      create = 
      Gui, 99: -Caption 
      Gui, 99:Margin, 0, 0 
      Gui, 99:Add, Edit, w500 r5 voutput 
      x := A_ScreenWidth - 500 
      Gui, 99:Show, NoActivate x%x% y0 
   } else 
      GuiControlGet, output, 99: 
   GuiControl, 99: , output, %output%%str% 
   Gui, 99:Show, NoActivate 
} 
PrintLn(str) { 
   Print(str . "`n") 
}

[Xander]

I've put documenting it on my Todo list. Probably would be best as the source (Java code I wrote quite awhile ago) wasn't documented at all either. Well actually I never really finished the Java version because I meant for it to recognize all characters and this method is not good for that. But it works fine for recognizing a subset of characters.

About the PrintLn(str) function... I don't understand your assigning str statement. str is just the parameter passed to the function. PrintLn(signature) is a line that is commented out in the ODR.ahk file.

The signature of what is read works like this:

First the first few columns are scanned and the number and position of of pixels are marked (depending on the size, sometimes it only scans the first column). For example, the number 7 has the following signature for the first column: 2AC. 2 means it has 2 hits, A tells you the first hit starts at the top of the character and ends somewhere in the middle. C signifies that the hit starts somewhere in the middle and ends at the bottom of the character. Similarly B would signify that the hit starts in the middle and ends in the middle, and D would signify that the hit starts at the top and ends at the bottom. Then the last column is scanned and that signature is appended to the previous, and similarly the top row and bottom row are scanned in that order.

So the number 7 has the signature of 2AC1A1D1A. Another font of 7 may also have the signature of 1A1A1D1B so both of those have to be accounted for...

Hope that helps for now.

[Xander]

Documentation has now been included. Also here is an example:

#Include ODR.ahk
#NoEnv

Gui, +LastFound
Gui, Add, Text, w100 HwndTextID, $123,456.7890
Gui, Show
ControlGetPos, x, y, w, h, , ahk_id %TextID%
text := ODR_ReadArea(x, y, w, h)
MsgBox, Read text area as: %text%

[Xander]

Impressing. . My outcome:
Displayed: $123,456.7890
Detected: $ 23 45 .7830

You need to update to the newest version, that was what I got with the example when I first tried, but updated the mapping to support the font used in the example :wink: Though I believe the comma was read for me.

[AHKisNice]

Is it possible to break a youtube captcha?

[majkinetor]

Interesting, but not very useful the way it is now. Just adding

Gui,  Font, s12

infront of your example, will make results wrong and changing the font size is typical operation.

[Moki]

This might be a stupid question but how do I use this to read a simple numbered only typed captcha on a website

[Moki]

bump ... still interested

[lrh9]

I was actually working on the same thing.

<!-- m -->http://www.autohotke... ... highlight=<!-- m -->

No dice with the ocr. It was too inaccurate and slow.

I suspect ImageSearch will be too slow as well.

The next thing I would try to do is use the dllcall readprocessmemory to obtain the information, but I don't know where in the pokerstars memory information is at.

Or you could try to reverse engineer the dlls from a preexisting poker utility that features auto reading of cards.