Spy Date Field issue

[Tejaskumar_Darji]

Hello team,

I've to spy below date field and it has DDMMYYYY written in background in light grey color, and on top of that we have date with bold black font.

Decipher is currently grabbing mix data e,g. 0M 2M 00 D1 2Y 0Y 2Y Y1 etc etc



Is there any solution to avoid this thing in decipher?


------------------------------
Thanks & Regards,
Tejaskumar Darji
------------------------------

[shashank.kumar280]

Hi Tejaskumar,

You could also do a regex operation on the output text "0M 2M 00 D1 2Y 0Y 2Y Y1" to filter out only the numeric values. There are some regex utilities available on DX.

------------------------------
Shashank Kumar
DX Integrations Partner Consultant
Blue Prism
Singapore
+6581326707
------------------------------

[PatrickAucoin]

I see two areas where an OCR engine could potentially struggle. First is DDMMYYYY om grey, similar to a watermark, that you want to ignore. Second is the "constrained print field," the comb-style lines on the form that divide each character in the field.

My first inclination would be to try to adjust the thresholding to filter out or ignore the grey background text, but I don't think Decipher lets you do that for now.

Here is what I would try: instead of defining a single field to extract the entire date, define 8 individual fields, and validate each field as a single digit. You could then combine the 8 characters into a new field for downstream processing.

------------------------------
Patrick Aucoin
Senior Product Consultant
Blue Prism
------------------------------

[HarpreetKaur]

If you can ensure that the behavior is thoroughly consistent, you could potentially add a calculation stage in your blueprism solution to remove M's, D's and Y's in your date data item.

------------------------------
Harpreet Kaur Product Consultant
------------------------------

[Ben.Lyons1]

Hi Tejaskumar,

A number of great suggestions already and worth trying each of these to see which yield the best results. I would also consider adding "IcrDigit=true" to the Misc. Function field as this should limit the result to numbers only and ignore any letters.

Thanks

------------------------------
Ben Lyons
Product Consultant
Blue Prism
UK
------------------------------

[Tejaskumar_Darji]

Hi Harpreet,

Yes, that is a good idea to handle it downstream but currently, behavior is not consistent with Decipher.

------------------------------
Thanks & Regards,
Tejaskumar Darji
------------------------------