Jump to content
Sign in to follow this  
nbg15

tesseract doesnt detect the easiest image *image to text*

Recommended Posts

nbg15

Hello everybody..

 

i have this picture here *attached* and this script here: GDIPlus_Image2.jpg

 

$ImageToReadPath = @MyDocumentsDir & "\GDIPlus_Image2.jpg"
$ResultTextPath = @MyDocumentsDir & "\Result"
$OutPutPath = $ResultTextPath & ".txt"
$TesseractExePath = @MyDocumentsDir & "\Tesseract.exe"

ShellExecuteWait($TesseractExePath, '"' & $ImageToReadPath & '" "' & $ResultTextPath & '"', "", "", @SW_HIDE)


If @error Then
    Exit MsgBox(0, "Error", @error)
EndIf

MsgBox(0, "Result", FileRead($OutPutPath))

FileDelete($OutPutPath)

 

but tesseract doesnt recognized the correct word... and gives me trash back...

this is the image >> GDIPlus_Image2.jpg

and the result was >> "samm:lmao:

the image was an normal jpg and generated with this code here:
 

_ScreenCapture_Capture(@MyDocumentsDir & "\GDIPlus_Image2.jpg", 712,268,853,284)


Could anybody give me a hint what i can do better to get this easy image to text?

 

thank u very much!!!

 

 

Edit: i also tried to capture the screen as bmp with a higher resolution... nothing changed... 

 

 

_ScreenCapture_SetBMPFormat(4)
    _ScreenCapture_Capture(@MyDocumentsDir & "\GDIPlus_Image.bmp", 712,279,853,295)

 

Edited by nbg15

Share this post


Link to post
Share on other sites
LarsJ

You can open the image in Paint, double the width and height, save the image and try again.

It's a small image, and there is a risk that there simply are too few pixels to recognize the characters. It usually helps to make the image larger.

Share this post


Link to post
Share on other sites
Pain

Hi nbg15,

You should be able to resize your image with this script:

#include <GDIPlus.au3>

; init
_GDIPlus_Startup()

; get file
$sFile = FileOpenDialog("Select an image (non JPG)", "", "Images (*.bmp;*.png;*.gif;*.tif)")
$hImage = _GDIPlus_ImageLoadFromFile($sFile)

; size
$width = _GDIPlus_ImageGetWidth($hImage)
$height = _GDIPlus_ImageGetHeight($hImage)

; resize image 200%
$hBitmap_Scaled = _GDIPlus_ImageResize($hImage, $width * 2, $height * 2)
; save file
_GDIPlus_ImageSaveToFile($hBitmap_Scaled, @ScriptDir & "\resized.png")

; clean up
_GDIPlus_ImageDispose($hImage)
_GDIPlus_BitmapDispose($hBitmap_Scaled)

; done
_GDIPlus_Shutdown()

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this  

  • Similar Content

    • Xandy
      By Xandy
      Special thanks: AdmiralAlkex, Melba23, MrCrearoR, Dragon Warrior 3, SDL
      MapIt is a tile world editor.  MapIt was built around the concept of reversing Dragon Warrior map images.  MapIt can take image input and produce a tile and world array.  
      Changing and replacing tile / world data is easy.  B/c tile world editor.

      CTRL+R in image above to signal replace tile action and I use "G" to Get the tile under mouse.
      A full list of hotkeys can be assigned in the: Help Menu\Hotkeys
       
      MapParser is a C++ project that scans images for unique tiles. 
      MapIt can be downloaded without MapParser. MapParser can be toggled off in the Scan_Image dialog. Without MapParser, MapIt will use the Scan_Tiles() function written in AutoIt ; which is 100 * slower Idk. If MapParser.exe will not run for you:      Installing Visual C++ Redistributable for Visual Studio 2015 should fix it: https://www.microsoft.com/en-us/download/details.aspx?id=48145   
      You can start with example world and tiles.
      Example world was made following these steps:
      Started with a tile map image of DragonWarrior3 town of: Reeve From MapIt World Menu \ New \ Scan_Image dialog, I set the area to exclude the key legend to the far right of image. After scanning the map image to world and tile array.  I removed a few of the map artifacts.  More work could be done on this world; removing unwanted tiles, but it is fine for now. I saved my world to disk.  This creates folder: Worldname: Containing folder of Tiles and a Worldname.txt. Using The Gimp, I edited some tiles to have a transparent color: Stairs, Trees, Desk Tables, Chest-of-drawers, Chairs, Signs, Doors, Beds. I changed the world layers to 2: World Menu \ Properties (*I recently changed this path) F9 Finds all selected tile on current layer and changes to a new selected tile on new layer. I used F9 to change all Trees on layer: 0 to Trees on layer: 1. Then I used F9 to change all Trees on layer: 0 to Grass on layer: 0  
      In this video you can see how I used the Tile Menu \ Replace From Disk option to remap tile images to my custom tiles.  Conveniently my tiles already have a transparent pixel.
      See video for how that was done:
      To use the example world:
      First unzip the world save file: http://songersoft.com/programming/mapit/worlds/Reeve_Swapped.zip From the World Menu: choose \Load Navigate to the Reeve_Swapped.txt located in the extracted zip. Or you can scan any image.
      The map images I used are here: http://www.realmofdarkness.net/dq/games/nes/dw3/maps/world
      For download, videos, and example of created world file data; please visit the MapIt webpage: http://songersoft.com/programming/mapit/mapit_about.phtml
    • PuneetTewani
      By PuneetTewani
      #include <IE.au3>
      #include <Tesseract.au3>
      #include <MsgBoxConstants.au3>
      #include <Math.au3>
      #include <FileConstants.au3>
      #include <StringConstants.au3>
      #include <File.au3>
      #include <ScreenCapture.au3>
      #include <sound.au3>
      #Include <WinAPI.au3>
      #include <Date.au3>
       
      $OCR_Result = _TesseractScreenCapture(0,"",1,2,220,660,500,730,1)
      $OCR_Result1 = _TesseractScreenCapture(0,"",1,2,220,660,500,730,1)
      $OCR_Result2 = _TesseractScreenCapture(0,"",1,2,220,660,500,730,1)
      $OCR_Result3 = _TesseractScreenCapture(0,"",1,2,220,660,500,730,1)

      $sound = _SoundStatus("C:\ExpertAdvisorBuyAlert.wav")
      while _nowtime < 3.30 pm
          If $sound = True Then
             if $OCR_Result1 > $OCR_Result2
             
          EndIf
      EndIf
      Wend
      Trying to ocr some values on chart in real time(once per minute) and buy/sell securities on basis of alert generated in my software.
      I am struck onto few steps.
      1. On Tesseract Screen Capture indentation parameters. How can we determine the exact parameters if I just want numeric values only.
      2. The Tesseract Screen Capture generates and error Obj1 on line 185 which needs to be resolved.
      3. Sometimes lines get overlapped with values. What to do in that case.
      3. Detecting the sound as and when it approaches and then comparing the ocr values to decide on either buy or sell.
      The values that needs to be fetched are encircled.

    • reida
      By reida
      Hello. On page there is mail address. How to keep in a text file this address? 
      It is page and email
      https://alpari.com/
      info@alpari.com
    • caramen
      By caramen
      Hey.  
       
      I requested help about how to get a value from a text in a variable. 
      Now i know how to do that. But i learned with the command FileRead. Now i whould like to know how to replace the command :
      FileRead('Dossier.txt') The purpose is to read a webpage text. To find some value inside. 
       
      Btw i tryed to play with WindowsInfo.au3 but i dont got much thing.  
    • Atoxis
      By Atoxis
      Howdy, I've gone through a lot of au3 forums, and I once had a working Imagesearch script that I got from here.  However, and i'm just totally not sure how but my imagesearch scripts aren't working anymore.
      I'm not new to au3 but i'm not the most experienced with it's syntax/commands.

      Anyways, I've looked over the big threads involving imagesearch.

      Does anyone have a working Imagesearch x64 for win10 that is currently working as of the date with the post.

      Dll's and what not is fine, just when I tell the script to run, I want to be able to find the image on the screen!
      Can't find a working copy so if anyone has one please send it my way lol.

      I've taken all the imagesearch downloads and what not and have played with them but I can't get any of them working on my end, despite others saying they're working.
      Thanks.
×