Scanning Tips For Thin Paper

[Adam B.]

I'm trying to scan a book I own. I have a sheetfed document scanner at work that can scan to many different formats and perform OCR on it.

I've scanned it as a black and white document at 300dpi. I'm not very impressed with the quality (see screenshot). The illustrations don't look very good, and there are random black dots around the page. In addition, the text doesn't look very clean.

There are quite a few illustrations and footnotes, so converting it to a text document will loose a lot of the "magic" of the book.

I have a plethora of options and formats I can save it into directly from the scanner (PDF, TIFF, etc). The pages are also very thin, so if I scan it as a color document (and greyscale, probably), the back side will bleed through.

Has anyone else scanned a book like this with good results? What settings/format did you use?

[Elfwreck]

To make a good copy, I suppose I'd scan the whole thing as 300 or 400 dpi tiff, B&W, first--and then scan the images as color or greyscale as appropriate. I'd use the B&W pages to OCR & get text, and insert the color/greyscale pictures during formatting.

If I got bleedthrough on the pictures, and I cared to spend the effort (and if that's the book, it's worth the effort), I'd tinker with them in Photoshop. (Or find out if the author would consider releasing the original images in some kind of digital format, which is possible.)

[DDHarriman]

Dots on white paper indicate you have too much contrast, lower it and/or play with the gama settings of your scanner.

Scanning greyscale will also lower the situations you report and also give better OCR results sometimes.

What are you pretending to get: a OCRed PDF file? and if so, a image over text or a text and images at the same level?
Or another type of result?

[ath]

Originally Posted by Adam B.

I've scanned it as a black and white document at 300dpi. I'm not very impressed with the quality (see screenshot). The illustrations don't look very good, and there are random black dots around the page. In addition, the text doesn't look very clean.

What scanner are you using? What kind of scanning is it intended for? If it's a photo scanner, you should probably not try bilevel scans at 300 dpi.

Anyway, do not let the scanner do the processing (in your case contrasting and thresholding). Instead you scan, say, in greyscale, and then separately do whatever smoothing and sharpening you need before you reduce sampling rate and final thresholding (Photoshop, PaintShop Pro, etc.). That way you avoid scanner firmware problems.

I have a plethora of options and formats I can save it into directly from the scanner (PDF, TIFF, etc). The pages are also very thin, so if I scan it as a color document (and greyscale, probably), the back side will bleed through.

The best approach for thin pages is to insert black paper 'behind' the scanned page -- that lessens any reflected light picking up the printing on its other side. It takes a lot of extra work, but if you want high quality you won't mind working a bit.

Has anyone else scanned a book like this with good results? What settings/format did you use?

I don't think settings are relevant without knowing what scanner you have.

[Adam B.]

I'm using a Kodak Scan Station 100.

I think that scanning in greyscale will give me the best options. I did a few tests with greyscale and 400 dpi, and the result looks good. I'll have to remove the grey in the background (and inside the letters, time to polish my photoshop skills). The only downside is file size, but I can work on that once I have a good quality scan in place.