Is it possible to add a parameter for numeric-only OCR recognition?

[XMB-7]

When recognizing numeric-only content (e.g., verification codes, tabular data, serial numbers), misidentifications of similar characters (e.g., 0/O, 1/I, 5/S, 8/B) sometimes occur, so I wonder if a Boolean parameter like digit_only can be supported in the option.

[Descolada]

The OCR engine itself doesn't support recognizing only digits, meaning if such an option would be implemented then it would have to be done after doing the OCR, basically post-processing. Unfortunately I don't think a reliable solution is possible, as there will always be ambiguous cases, for example B could represent 8 or 13. Additionally I don't think I could cover all characters which could be digits, meaning it could fail with certain input (or the library would throw an error?).

I should note that other OCR engines such as Tesseract support recognizing only digits.

[meyverick]

I have the same issue and thought it would exist because many time it wrongly find "f" instead of "1" or even "S" instead of 5.