I have a client who needs to read account numbers off of checks, but because the font used for the account number is so different, FlexiCapture isn't able to read these digits accurately. He's tried using every OCR text type in the Recognition tab of the account number field, and none of them work.
Would creating a User Pattern library be the only way to get these account numbers recognized accurately? Also, is FineReader Pro the only software that can create create User Pattern libraries? I'll need to let the client know if he needs to purchase additional software.
Thanks,
Ben
Would creating a User Pattern library be the only way to get these account numbers recognized accurately? Also, is FineReader Pro the only software that can create create User Pattern libraries? I'll need to let the client know if he needs to purchase additional software.
Thanks,
Ben
Comments
2 comments
Sometimes setting other recognition options like field marking type and/or despeckling image may help to improve the result. If none other options help, then yes, your customer could try training user pattern in FineReader Pro. But anyway there is no guarantee of getting appropriate recognition even after attaching trained pattern to FC as the result depends much on the images (quality and font) that you process.
However it is hard to advise without seeing the image sample and current options set for the problem field in the document definition. Could you please show us the image being processed (or at least its part) together with field recognition options?
Despeckling the check image seemed to solve the problem. Thanks for the information though!
Ben
Please sign in to leave a comment.