Question

Using Zone OCR for Document Splitting and Typing

Version 9 Scanning Quick Fields

Updated April 23, 2014

asked on April 15, 2014

Folks,

I have a series of medical records that are broken into distinct sections, but we've found the most expeditious way of scanning them is to combine them. Now I'm trying to use Zone OCR to break the sections into their respective record types by adding a Page that indicates a section break. It needs to be something unique, so I started with

*** [Lab Results] ***

*** [USOB] ***

*** [History and Physical] ***

I'm wondering, first of all, if this is the best method. I'm also experiencing some issues with OCRing if the text is slight askew despite having added deskew, and I wondering if someone hasn't done some experimentation about the most efficient way to "read" this data.

Thoughts:

Ex. Adding two sets of lines to assist the deskew process "------------------------"

Ex. Changing to a shorter and more consistent section indicator (4-digit section code preceded by some special characters rather than letters).

I'd appreciate hearing from someone who has done this already cracked this nut.

Thanks,

Adam

0 0

Replies

replied on April 15, 2014

Have you thought about using a barcode? Deskew works pretty good on them and I've not had any issues. We're even using them on envelopes which don't always run through straight.

0 0

replied on April 15, 2014

I would really recommend the use of barcodes if possible. The other option is to use OMR after the preclassification and identification processes return the new document. Use conditional sequences to indicate which Zone OCR you may need for whichever document you are using as indicated by the barcode or OMR

0 0

replied on April 15, 2014

Thanks Kenneth, I don't own a Barcode license. Most of what we are doing is Zone OCR, so I'm making due with the license I have. (Probably should have mentioned that. I also don't have a document typing license.)

0 0

replied on April 15, 2014

You could see if it helps to do a full page OCR with a local deskew and then tell the Zone OCR process to use existing text. It will take a little longer, but sometimes doing operations in different orders helps out some.

0 0

replied on April 15, 2014

Hey Adam,

I've definitely used the Zone OCR before to separate different pages so what you're suggesting should work. It's hard to tell for sure if Zone OCR is the best activity to use since I don't know what your documents look like. The accuracy of the Zone OCR also depends on the quality of the scan. So you will want to make sure that you are scanning in the cleanest copies of those medical records as possible. Additionally, I would go for a shorter and more consistent section indicator as you suggested since I've ran into troubles with the * characters before with the OCR. Text and numbers tend to do better in the OCR based on my experience.

With deskew, try changing the settings around with the lines and text to see if it gives you different results. Sometimes there are some pages that just won't deskew 100%

As a side note, I've also seen people use self generated barcodes to help with the separation process. Barcodes tend to be more accurate than the zone OCR.

0 0

replied on April 23, 2014

Hi Adam,

It looks like you have a number of good answers to your question. If one provided the answer you needed, please click the “This answered my question" button.

If you still need assistance with this matter, just update this thread. Thanks!

0 0

You are not allowed to follow up in this post.

Question

Question

Using Zone OCR for Document Splitting and Typing

Replies

Sign in to reply to this post.