Currently, we are doing some pre-processings and use tesseract to extract text from this particular receipt.

Please note that I’m only asking for this particular receipt because I hope it’s easier to answer the question with a specific problem and avoid being generic.

Here’s the original receipt:

After pre-processing, the image looks like this:

The pre-processing involves blurring and etc. The result is quite bad, so I’ll skip it.

The extracted text using Tesseract is jibberish as expected because the pre-processed image is so bad. Here is one part:

™7 174 Chse CRYVALL SORHS 515 iy X 7
o2t 130 Ll PLST 1K 1A
JBTOTAL 763.25 AT
SALES TAX 78.23 L SRy
Torak $841.48 T :

I wonder how you would pre-process this image, so the OCR would result in a better accuracy and precision.

Thank you so much.