Question

Zone OCR in QF 9 Used to Identify and Populate Metadata

Quick Fields

Updated July 23, 2014

asked on July 2, 2014

Hello,

Okay, I have asked this question before, but this time I want to add/modify question. I am in the process of scanning 22 year old workman comp files. Within each document there are different types of pages--some same like medical claims--subpoenas, hand written notes, doctor notes, etc.. So, I have tried setting up my session as follows: 1. Look, identify the start of this document. See special text within this OCR Zone? Set conditions that "contain" and "equal" this text. 2. On this same sheet of paper (used as a cover/slip sheet to identify start of "new document", I have also created two OCR Zones to capture "text" to fill in my fields-- a. Employee Name and b. Employee ID No. under page processing. After the Zone OCR's, I have remove page 1. So, that makes three OCR Zones. When I test the processes, they work. When I actually scan these documents, they don't separate and my fields are full of garbage. I'm wondering because of all the videos, and white papers, etc. I've been watching/reading, that all the "papers" within the "document" have to be the same to run within the session? As in my CPP videos (working on Capture II Specialist) I've been watching, they show easy documents such as benefit checklists, applications, and they are all alike.. So, I'm thinking my problem might be that because all these pages are "different" that QF 9 isn't capable of this type of project. Should I be scanning these pages directly thru Laserfiche?

I have tired setting up this session with two sample pages--one First Page Identify "Scan New Document". Second sample sheet under page processing to "capture" my metadata to send to my fields. Doesn't work. The first sample page shows up in the second sample page place. Like I can't keep "them" there. Any suggestions? I have many files later that will be the same--pages within folder that will all look alike or many different types of pages. I am presently working with my supervisor to see if we can "tap" into our employee database and use the Lookup process, but they may not happen for a few months. Any suggestions? I'm at my wit's end and can't figure out what I am doing wrong. I'm on my own here. Thanks

Zone 2 and 3 All with test processes.PNG

My Session Set Up.PNG (11.17 KB)

| Download

My results.PNG (10.83 KB)

| Download

Zone 1 Identify.PNG (30.5 KB)

| Download

Zone 2 and 3 All with test processes.PNG (43.66 KB)

| Download

0 0

Replies

replied on July 3, 2014

Hi Susan,

Any chance you could upload a copy of the session you're having trouble with? And perhaps some sample documents? It's a little difficult to determine what the issue is without being able to poke around through the actual session itself.

Keep in mind, when you're configuring your zone OCR processes, I believe the default setting is to process page 1. Unless you change that manually, the process will always look on page 1, which doesn't have the information you're looking for.

0 0

replied on July 7, 2014

Thank you so much Brett for responding.

Set to read page 1. I sent some more pictures of this session. Okay, it is now identifying "New" document. See the Revision Panel of my Session PNG I have attached. Note while there the "document name".

Please look at Meta Data Fields. See the "garbage". When I test these processes before scanning they work. Zone OCR reads Doe, John.

Please note Session 1 set up out pane. Is it reading every page and taking information from last page and filling in metadata even though I have this set up to read first page only?

I can't send examples of the different pages within my documents because of HIPPA and other federal regulations. I'm pretty sure I have these Zone OCR set up correctly. Suggestions?

Session 1 set up.PNG (189.9 KB)

| Download

Revision Panel of my Session.PNG (25.05 KB)

| Download

Meta Data Fields.PNG (252.04 KB)

| Download

0 0

replied on July 11, 2014

Hi Susan,

Without having a copy of the actual QFX file and some sample documents, it's really difficult to deduce exactly what's going wrong. If possible, I would strongly recommend that you work with your reseller (perhaps with some dummy sample documents) to work out what's going wrong here. That said, I'll try to help out a little based on what I can see!

The OCR definitely seems to be grabbing information from the wrong page. I think what it's actually seeing is the text "ASSOCIATED GROCERS, INC." and "Released to FULL DUTY as of 6/10/14", but because the page is skewed, it's not able to get a good read. I think using a Deskew process (rather than Rotate) will at least help you to get readable data, even if it's from the wrong page.

I'm not sure that I can tell you why it's reading from the wrong page though. I see that there are 3 documents in the Document Revision pane, and none of them are named anything even close to each other. That would imply that your identification criteria may not be set up correctly. Is it possible that what you're thinking is page 1, is not actually being identified as page 1?

0 0

replied on July 23, 2014

Susan, will you export your session and upload the resulting QEX file so we can take a look at it?

0 0

You are not allowed to follow up in this post.

Question

Question

Zone OCR in QF 9 Used to Identify and Populate Metadata

Replies

Sign in to reply to this post.