Weird OCR results

I used Adobe Scan for Android to scan a numerical tabular document and then I saved it to a pdf file. The OCR results were unexpectedly poor, with several digits recognized as alphabetical or special characters.

The source photo was good and sharp, and some incorrectly recognized digits in one place were recognized correctly in other locations. The quality of the source digit in all the locations were of comparable quality and clarity.

What can be done to improve the quality of the results?

Perhaps a flag to indicate that the image is text only, numeric only, or mixed could be added in a drop-down option list to select before saving and OCRing the scan? The current selection would be shown on the drop-down button.