text distortion of scanned pages

Hi all,

I recently photographed some pages from a very old book. JPG format. Now, my character regognition program needs to try to read them but the pages are curved - not flat. (I was unable to hold them down flat as I snapped the shutter and held the lens cap out of the way. Sigh....)

Can Photoshop correct this curvature distortion, making the lines straight? How?

Thanks, Zane

----- Original Message -----
From: "Z A" <>
Newsgroups: alt.graphics.photoshop
Sent: Friday, February 13, 2004 1:05 AM
Subject: text distortion of scanned pages

Hi all,

I recently photographed some pages from a very old book. JPG format. Now,

my

character regognition program needs to try to read them but the pages are curved - not flat. (I was unable to hold them down flat as I snapped the shutter and held the lens cap out of the way. Sigh....)

Can Photoshop correct this curvature distortion, making the lines

straight?

How?

Thanks, Zane

I've scanned quite a few old books and almost daily scan old magazines. With the magazines I remove the staples separate the pages, it's so much easier.

With the old books it's preferable to possess one where the bindings are actually coming apart.
If it's a borrowed book, which you need to be careful with, there is not going to be much you can do. You take what the scanner delivers. If it's your book, buy a piece of black cloth, cover the book and then set another book of similar weight on top of the book you scanning. In most instances the center-fold will go right down, producing a good scan.

JPG's for OCR?
Not sure that's possible. Most OCR requires TIF files (line-art) when scanning.

I have four books which are over 100 years old and they scan quite well. Of course my purchase of these books was entirely for the data-content rather than to retain the appearance and/or condition of the book. One of them, I actually took the binding apart intentionally because of the number of pages.

On Fri, 13 Feb 2004 3:37:03 -0500, lostinspace wrote
(in message <P00Xb.2169$>):

JPG's for OCR?
Not sure that's possible. Most OCR requires TIF files (line-art) when scanning.

I have four books which are over 100 years old and they scan quite well. Of course my purchase of these books was entirely for the data-content rather than to retain the appearance and/or condition of the book. One of them, I actually took the binding apart intentionally because of the number of pages.

Thank you for the prompt reply.

Assuming that the JPG's can be converted to something readable, can Photoshop 7 "uncurve" the pages. How? The OCR reader prefers straight lines.

The archivist would not allow the old books to be laid on the copy machine for the reasons you mentioned. Thanks.

Zane Anderson

"Assuming that the JPG's can be converted to something readable, can Photoshop
7 "uncurve" the pages. How? The OCR reader prefers straight lines."

Not that I'm aware of.

Your best possibility is to get a LASER print of the images you have and then scan and edit from the print.

"Z A" wrote in message

...
character regognition program needs to try to read them but the pages are curved - not flat. (I was unable to hold them down flat as I snapped the shutter and held the lens cap out of the way. Sigh....)

Can Photoshop correct this curvature distortion, making the lines

straight?

How?

It should be possible using the Shear filter, but it would be tedious. It might be possible using the Liquify filter, but I've never tried Liquify for transforming an entire image in a uniform way.