Can you OCR the images inside of .pdf files? - Page 3

mrmikel · 09-13-2014, 08:49 AM

If you get yourself on the email list for readiris or serif, versions of paid ocr programs are offered at what seem to me reasonable prices.

As has been noted, there are no free alternatives that do 1/10 of what the commercial ones do. This may be because there is real money in doing document conversion, so no one is willing to do for free what customers are willing to pay for. Also support for these programs must be a headache. At best OCR produces a fair number of errors in the output unless the input is perfect and simple. The complaints about this may be a real deterrent to a software author.

Hitch · 09-13-2014, 03:56 PM

Quote:

Originally Posted by mrmikel

If you get yourself on the email list for readiris or serif, versions of paid ocr programs are offered at what seem to me reasonable prices.

As has been noted, there are no free alternatives that do 1/10 of what the commercial ones do. This may be because there is real money in doing document conversion, so no one is willing to do for free what customers are willing to pay for. Also support for these programs must be a headache. At best OCR produces a fair number of errors in the output unless the input is perfect and simple. The complaints about this may be a real deterrent to a software author.

mrmikel:

Not a lot of money. All of the better scan houses I know are being run out of business by the super-cheapie, low-quality places, (which shall remain nameless). Cheap, in this day and age (or better yet, FREE FREE FREE!!!) is the mantra.

It's a shame, because trust me: we get a LOT of scanned/OCR in here, and the freebs or cheaps really suck. I just had to quote a guy nearly a thou, to clean up and format three books, because HE scanned them (to save Money, of course!!!). Found this "wonderful machine" that just "made my books into Word files." Uh, yeah. Replete with manually-retained running headers and a BAJILLION broken paragraphs. I'm hoping that Tox's ePUBTools will help us survive it. Those books would have been hundreds--not a thou--if we'd gotten Abby-output instead.

The downside of the digital age is manifest; everyone thinks that typing makes them authors; that using Word makes them layout and eBook-conversion experts; that being able to put a PDF into a freebie website to convert into Word makes them Scanning/OCR experts. Oh, and having written 10 reviews on Amazon makes one a qualified Book Editor.

But, bygod, it's FREE FREE FREE!!!!

Hitch

BetterRed · 09-13-2014, 06:47 PM

- Evernoted

Hitch · 09-13-2014, 06:51 PM

Quote:

Originally Posted by BetterRed

- Evernoted

Wellll....

Gosh, THANKS!!! That's all I know to say to that. :-)

Hitch

rkomar · 09-13-2014, 07:30 PM

It's possible to get more information from tesseract by specifying HTML output (hOCR) rather than straight text. Tesseract is great, but requires a fair bit of work from the user to tune it properly. It's definitely not a point'n'grunt solution.

mrmikel · 09-13-2014, 09:21 PM

A lot of people don't know what they don't know, I guess. I am fortunate my customers mostly know when to quit.

For the rest I give them an option to mess with it themselves. If they are lucky, maybe they save a buck or two. If not, I tell I have to not only figure what was wrong in the first place, but repair their mess as well. As you noted, this increases the cost massively. I suggest I look forward to them paying for a trip to Tahiti for me.

You have the disadvantage of trying to compete with free, free, free. Just a demonstration that on the internet nobody knows you're a &^*&^&*^&.

A friend of mine sent me a reminder:

We offer three kinds of work, good, cheap, quick. You can only have two at once.

A cheap, good job will not be quick.
A quick, good job will not be cheap.
A cheap, quick job will not be good.

I was thinking on the conversion end of document scanning for big companies so they can get rid of paper altogether.

markom · 09-14-2014, 07:54 AM

Quote:

Originally Posted by mrmikel

...

I was thinking on the conversion end of document scanning for big companies so they can get rid of paper altogether.

They can get rid of paper without the need for scan to epub/mobi conversion though, by reading OCR-ed A4/letter sized pdfs on DPT-S1 (there will be a lot cheaper 13" e-ink readers sooner or later) or some good 10" e-ink/lcd tablet in landscape or 12" tablet/netbook if they don't mind the weight.

mrmikel · 09-14-2014, 08:59 AM

Right on. It was the conversion end I was thinking of. PDFs are a much better choice for that sort of thing since you can maintain the original pretty much as you got it while doing enough of a text conversion for indexing.

klmmc13 · 10-16-2014, 01:20 PM

Well Gentlepersons, I'll be able soon to follow your recommendations and spare my pocketbook at the same time. Come to find out that my sister had acquired a copy of AAbby finereader and no longer uses it, and will be passing it along to me.

Thank you all for your advice and assistance.
Kathy the MamaDragon

grumbles · 10-30-2014, 08:07 PM

I didn't attempt to wade through all the previous noise in this thread but this is how I have converted pdf files to epubs.

I use PDFill (a free Windows program) to covert the PDF file to images (usually at 600 dpi).

I then run the images through ScanTailor to make the pages uniform, in particular I want all the headings and/or page numbers the same distance from the top and bottom edges. Where there are no headings/page numbers, I want make sure there is equivalent blank space.

I then use XNView to trim the headings and/or page numbers.

Then I run the OCR program on the images. I use Abby sprint that came with my scanner.

This works well and goes fairly quickly. I've used this on several PDF files I've downloaded from the Internet Archive.

09-13-2014, 06:47 PM	#33
BetterRed null operator (he/him) Posts: 20,553 Karma: 26954694 Join Date: Mar 2012 Location: Sydney Australia Device: none	- Evernoted Last edited by BetterRed; 09-13-2014 at 06:51 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
no text extraction for pdf with images and OCR	fxp33	Conversion	7	12-15-2015 07:22 AM
Cover images for pdf files on Kindle PW	blz777	Amazon Kindle	0	07-21-2013 10:45 AM
Google Adds OCR for PDF Files	kjk	News	0	06-22-2010 02:27 PM
Can I view images in PDF files ?	eisho	Sony Reader	1	08-03-2008 08:49 PM
Sony reader for PDF files: pages as images	claudioita	Sony Reader	3	07-30-2007 02:46 PM

09-13-2014, 08:49 AM	#31
mrmikel Color me gone Posts: 2,089 Karma: 1445295 Join Date: Apr 2008 Location: Central Oregon Coast Device: PRS-300	If you get yourself on the email list for readiris or serif, versions of paid ocr programs are offered at what seem to me reasonable prices. As has been noted, there are no free alternatives that do 1/10 of what the commercial ones do. This may be because there is real money in doing document conversion, so no one is willing to do for free what customers are willing to pay for. Also support for these programs must be a headache. At best OCR produces a fair number of errors in the output unless the input is perfect and simple. The complaints about this may be a real deterrent to a software author.

09-13-2014, 07:30 PM	#35
rkomar Wizard Posts: 2,985 Karma: 18343081 Join Date: Oct 2010 Location: Sudbury, ON, Canada Device: PRS-505, PB 902, PRS-T1, PB 623, PB 840, PB 633	It's possible to get more information from tesseract by specifying HTML output (hOCR) rather than straight text. Tesseract is great, but requires a fair bit of work from the user to tune it properly. It's definitely not a point'n'grunt solution.

09-13-2014, 09:21 PM	#36
mrmikel Color me gone Posts: 2,089 Karma: 1445295 Join Date: Apr 2008 Location: Central Oregon Coast Device: PRS-300	A lot of people don't know what they don't know, I guess. I am fortunate my customers mostly know when to quit. For the rest I give them an option to mess with it themselves. If they are lucky, maybe they save a buck or two. If not, I tell I have to not only figure what was wrong in the first place, but repair their mess as well. As you noted, this increases the cost massively. I suggest I look forward to them paying for a trip to Tahiti for me. You have the disadvantage of trying to compete with free, free, free. Just a demonstration that on the internet nobody knows you're a &^&^&^&. A friend of mine sent me a reminder: We offer three kinds of work, good, cheap, quick. You can only have two at once. A cheap, good job will not be quick. A quick, good job will not be cheap. A cheap, quick job will not be good. I was thinking on the conversion end of document scanning for big companies so they can get rid of paper altogether.

09-14-2014, 08:59 AM	#38
mrmikel Color me gone Posts: 2,089 Karma: 1445295 Join Date: Apr 2008 Location: Central Oregon Coast Device: PRS-300	Right on. It was the conversion end I was thinking of. PDFs are a much better choice for that sort of thing since you can maintain the original pretty much as you got it while doing enough of a text conversion for indexing.

10-16-2014, 01:20 PM	#39
klmmc13 formatting student Posts: 47 Karma: 38268 Join Date: Dec 2013 Location: South Arkansas, US Device: several models of Kindle; Several Android tablets & 3 Android phones.	Well Gentlepersons, I'll be able soon to follow your recommendations and spare my pocketbook at the same time. Come to find out that my sister had acquired a copy of AAbby finereader and no longer uses it, and will be passing it along to me. Thank you all for your advice and assistance. Kathy the MamaDragon

10-30-2014, 08:07 PM	#40
grumbles Addict Posts: 238 Karma: 1500000 Join Date: Nov 2009 Location: Toronto Device: Pandigital Novel (Black), T-2 and 3, Nexus 7	I didn't attempt to wade through all the previous noise in this thread but this is how I have converted pdf files to epubs. I use PDFill (a free Windows program) to covert the PDF file to images (usually at 600 dpi). I then run the images through ScanTailor to make the pages uniform, in particular I want all the headings and/or page numbers the same distance from the top and bottom edges. Where there are no headings/page numbers, I want make sure there is equivalent blank space. I then use XNView to trim the headings and/or page numbers. Then I run the OCR program on the images. I use Abby sprint that came with my scanner. This works well and goes fairly quickly. I've used this on several PDF files I've downloaded from the Internet Archive.

Advert

Advert