Community

Receipt Recognition Problem

Hi,

we are using processImage method to process receipts. The problem is that the text we get from OCR is imposible to parse.

Here i leave a file with 3 examples.

 

Do you have any recommendations for taking the pictures? We use the phone camera in our application.

Also, for parsing is better to use .txt or xml?

 

Thanks in Advance :)

 

Matias

0

Comments

1 comment

  • Avatar
    Nikolay Krivchanskiy

    Hi Matias,

    We managed to extract some text, using the following settings:

    ·         language=Spanish;

    ·         profile=textExtraction;

    ·         textType=normal,matrix;

    ·         imageSource=photo;

    The result of processing with these parameters attached.

    You can also try out our GUI example utility from downloads page, to try and process images using various settings, or processing them as a receipt to extract data from in automatically.

    For your issue you might be interested not in processImage method, but in processReceipt method.

    Also please make sure to check out the best practices section of our website.

     

    As for the parsing issue, the .xml is more convenient as it stores not only textual data, but also its location on the image.

    1
    Comment actions Permalink

Please sign in to leave a comment.