Notifications
Clear all

OCR software

16 Posts
10 Users
0 Reactions
3 Views
(@steve-owens)
Posts: 238
Registered
Topic starter
 

Would anyone care to recommend a decent OCR program?

I am particularly interested in something that can faithfully reproduce the columns of POINT-NORTHING-EASTING that we increasingly see in scanned .PDF plan sets for construction layout.

 
Posted : June 15, 2011 10:52 am
(@guest)
Posts: 1658
Registered
 

Have you tried the built in OCR that Adobe Acrobat has?

Personally I would rather contact the agency or developer and request actual data files prior to beginning the project. JRL

 
Posted : June 15, 2011 11:02 am
(@steve-owens)
Posts: 238
Registered
Topic starter
 

(1) Do not use Acrobat (a little pricey).

(2) Always the best way, but some folks are not forthcoming with CAD files.

 
Posted : June 15, 2011 11:19 am
(@jimcox)
Posts: 1951
Registered
 

Omnipage has worked well for me...

 
Posted : June 15, 2011 11:32 am
(@joe-m)
Posts: 429
Registered
 

I hear Google Docs will do OCR on a pdf file. I also know that many pdf files that are created directly from a digital format (rather than a scan) have text that you can copy and paste using the freely available Adobe Reader.

 
Posted : June 15, 2011 11:33 am
(@steve-owens)
Posts: 238
Registered
Topic starter
 

I gave that a shot and got baffled and mystified. Very odd for Google. I may have another go at it...

 
Posted : June 15, 2011 11:37 am
(@steve-owens)
Posts: 238
Registered
Topic starter
 

Doesn't look as if out operating system meets the criteria (sigh....)

 
Posted : June 15, 2011 11:39 am
(@blakehuff)
Posts: 491
 

There are a number of free online OCR sites.
http://www.onlineocr.net/
http://www.newocr.com/
I'm not sure how well it work with the columns but maybe you can import space delimited or find/replace spaces with comma and csv import that. I've done something similar in the past with decent results. I haven't had any luck with coordinate tables with column/row lines.

 
Posted : June 15, 2011 12:26 pm
(@a-harris)
Posts: 8761
 

Most scanners come with a control center program that will do OCR

The Brother MFC6490CW came with a decent routine.

It does not do the columns just like on the original, yet most point files are either comma or space delimited and it seems that with any OCR that some editing is needed.

 
Posted : June 15, 2011 12:46 pm
(@carl-b-correll)
Posts: 1910
 

> The Brother MFC6490CW came with a decent routine.
>
> It does not do the columns just like on the original, yet most point files are either comma or space delimited and it seems that with any OCR that some editing is needed.

Just the other day I tried to you my Brother MFC6490CW to scan in some old 2.5" TI coordinate tapes my Dad had printed on in the early 90's, and had mixed results... I got maybe 55% of what was intended, but ink blobs and many the "strike through" zeros came out as "8".

I think a lot of it is going to depend on the size and the font of the item you are working from if you use the Brother.

Just my $0.02.

 
Posted : June 15, 2011 1:09 pm
(@guest)
Posts: 1658
Registered
 

I can see being stingy with the CAD files. However an Ascii coordinate file is so easy to export and e-mail these days. Probably all boils down to who you are talking to. JRL

 
Posted : June 15, 2011 1:34 pm
(@steve-owens)
Posts: 238
Registered
Topic starter
 

Blake

I tried Onlne OCR with a multi-column-formatted list of points and coordinates AND it had dividing lines between the fields.

It did an impressive job!! Not 100%, but pretty close....

Thanks for the links.

 
Posted : June 15, 2011 1:48 pm
 RADU
(@radu)
Posts: 1091
Registered
 

OCR software came with my trusty old scanner. Actually the survey story that I sent you was scanned from type written copy and had limited editing.

RADU

 
Posted : June 15, 2011 1:57 pm
(@Anonymous)
Posts: 0
Guest
 

Steve I've used OCR software on and off for a number of years and have always been very wary of the results. But also pleased to ahve to not type heaps of figures.

One needs a careful check of the ouput especially dealing with coords.
You can't run a spellcheck 😛 as in word doc.

I'm not wanting to put you off, just saying don't take the results as gospel.

(Edited)
I get a lot of plans that are coordinated lists and in most cases can get an ASCII file sent.
Some need a bit of a massage but it beats other ways of inputting.
Explaining the need for accuracy and cost saving is a legitimate 'excuse' to have them presented that way.
I once asked an engineering firm that refused how they would deal with a contour and detail survey offered to them in paper form. It didn't change their mind but I bet it showed how pedantic one can get.

Richard

 
Posted : June 15, 2011 11:59 pm
(@steve-owens)
Posts: 238
Registered
Topic starter
 

Richard

In re: beware of the results - but of course!!

All I am looking for is OCR that is faster and more accurate than me - not a great demand, believe me.

Checking 500 coordinate pairs against the original is WAY easier than typing them in.

 
Posted : June 16, 2011 4:10 am
(@marc-anderson)
Posts: 457
Registered
 

PdfOcr has worked really well for me. It's not perfect, but it's very good, much better than most I've seen. Reasonably priced too.

 
Posted : June 16, 2011 6:44 am