As the foundation for the rest of the digital library process, doing a scan correctly can save hours of work down the line and keep a book from getting caught in a snag that forces it to spend years in the pipeline, potentially never seeing the light of day. Many of the later issues in the pipeline can be avoided if attention is given to scanning.
Before approving a scan to be passed on to the next step, certain things should be checked:
• Pages: Count every individual page in the pdf and make sure they are all there. Even if your number matches, you could have accidentally scanned one page twice and missed another one entirely. You need to check each page individually.
• Alignment: Are the pages straight? Pages that have been scanned in at an angle can easily get mangled in the OCR process, causing many issues later on. If there are loose-leaf pages, they can cover the text of other pages as well (See example A below).
• Contrast: Is there a high enough contrast on each page? Words from a low contrast page are often mangled by the OCR process, since the computer isn’t able to recognize them correctly. This causes many more hours of work at both the Proof 1 and the Proof 2 stage.
• Shadows: Are there lots of shadows on the page, particularly at the bottom of the page, in the margins? (See example B below). This can cause skipped or mangled words and lots of eyestrain for later individuals squinting wearily at the shadows and hoping they’ve correctly guessed at whatever words lurk there.
If there are problems with any of the above, it is worth it to take the time to redo those pages, or even the entire scan.
After checking all of the above, make sure the scan has the correct book number and the corresponding name before passing it on.
Additionally, if you have scanned the book in so that it is in multiple pdfs/parts, the parts should be combined into one pdf before it goes through the OCR process.
Bad Scans
Characteristics of a Bad Scan:
• Pages are not straight (examples A and B)
• Text is covered (A)
• Margins are angled with dark areas (causing text to be missed by the OCR) (B)
• Low contrast (check your printer settings)
• Grainy (B)
• Cut off part of the page (C)
A.
B.
C.
Good Scans
Characteristics of Good Scans:
• Pages are straight (example D)
• Pages are flat with no tenting in the margins (D) (all words are easily recognized)
• High contrast (Example D) – this can be done on the printer settings
• Not grainy, whenever possible (It’s not always possible to avoid. However, a light level of graininess is usually fine.)
• Best scans: Only have one page of the book per pdf page (Example E). That way, if there are problems, they are more easily fixed by someone further down the line. (Note: Scans that have come in as two pages can actually be split into one page scans in OCR. But it works better if it’s done at the Scan stage.)
D.
E.