MobileRead Forums - View Single Post - how to convert a scanned page from a book (looks like photo of page) to clean text?

DSpider · 11-23-2012, 03:55 PM

My advice to the original poster of this thread is to watch a few tutorials on how to use Word, Acrobat, maybe InDesign and Illustrator, as well. The "Essential" libraries from lynda.com are excellent and I highly recommend them. Of course, there are also open-source alternatives (Adobe software is very expensive), such as LibreOffice, Inkscape, Scribus, GIMP, etc., which can do the job very well. I'm a fan of open-source software (Arch Linux is my primary OS), but I have to admit that Adobe has the flagship software for this industry.

My current workflow is to scan text in grayscale, at 300 dpi, JPG (~90% quality setting), and images (like the covers, pictures or other graphics) in colour, at 600 dpi, TIFF.

First I start with the graphics, using Photoshop to get the most out of them, and vectorize what I can by manually tracing them in Illustrator (again, the "Essential" training library from lynda.com should be enough). Then I run the scans through ABBYY FineReader, proofread it once, export as RTF, run a custom macro in Word 2010 SP1 that keeps only the bolds, italics, subscripts, superscripts and inserts the footnotes as in-line text (separated with tags, so I know where to place them later). This macro outputs a squeaky clean RTF, which I import into InDesign CS5.5 and start redoing the layout, based on styles. Here I sometimes use Scan Tailor, but just for 1:1 comparison when redoing the layout. Usually I batch rename the images from Scan Tailor to match the page number, so that it's easier to go back and forth.

And finally, I proofread again the final e-book, but this time on my e-reader, highlighting the parts that I may have missed or that just don't look right. As you can imagine, the quality is very high, but only after putting a lot of time into it. An e-book could take up to a month.

Good luck. Many people give up after the third book or so.

11-23-2012, 03:55 PM	#12
DSpider Evangelist Posts: 450 Karma: 343115 Join Date: Nov 2009 Location: Romania Device: PW2 2014	My advice to the original poster of this thread is to watch a few tutorials on how to use Word, Acrobat, maybe InDesign and Illustrator, as well. The "Essential" libraries from lynda.com are excellent and I highly recommend them. Of course, there are also open-source alternatives (Adobe software is very expensive), such as LibreOffice, Inkscape, Scribus, GIMP, etc., which can do the job very well. I'm a fan of open-source software (Arch Linux is my primary OS), but I have to admit that Adobe has the flagship software for this industry. My current workflow is to scan text in grayscale, at 300 dpi, JPG (~90% quality setting), and images (like the covers, pictures or other graphics) in colour, at 600 dpi, TIFF. First I start with the graphics, using Photoshop to get the most out of them, and vectorize what I can by manually tracing them in Illustrator (again, the "Essential" training library from lynda.com should be enough). Then I run the scans through ABBYY FineReader, proofread it once, export as RTF, run a custom macro in Word 2010 SP1 that keeps only the bolds, italics, subscripts, superscripts and inserts the footnotes as in-line text (separated with tags, so I know where to place them later). This macro outputs a squeaky clean RTF, which I import into InDesign CS5.5 and start redoing the layout, based on styles. Here I sometimes use Scan Tailor, but just for 1:1 comparison when redoing the layout. Usually I batch rename the images from Scan Tailor to match the page number, so that it's easier to go back and forth. And finally, I proofread again the final e-book, but this time on my e-reader, highlighting the parts that I may have missed or that just don't look right. As you can imagine, the quality is very high, but only after putting a lot of time into it. An e-book could take up to a month. Good luck. Many people give up after the third book or so.