Alternate Format - Fixing poor quality scans for electronic reading.
As an Adaptive Technologist (AT for short - Important: not to be confused as IT) I spend a portion of my job creating what we call Alternate Format, or what can also be called electronic format. This is a process by which a physical document or book is scanned and/or converted into a format that is accessible electronically.Â
Working with publishers to get a good copy of the original text is the best way to start this process, and more and more it is becoming the norm. However, there are some times when, for what ever reason, the text has to be scanned in manually. Depending on the individual performing the scanning, or the quality of the original, we still sometimes end up with almost unusable results.
With poor quality electronic scans, our OCR (Optical Character Recognition) software has a much more difficult time creating live text, thus making electron reading difficult or impossible.Â
What I want to share here today is a method I recently discovered that can help increase the success rate of running an OCR of a poorly scanned or faded document. This process will require Adobe Acrobat Pro, and Photoshop CS6. This is the software I used, but it may work with an older version of these Adobe products.
Step 1: Open the faded PDF file in Adobe Pro. Once opened, select File > Save As > Image > and TIFF.
This will create individual graphic TIFF files of each page.
Step 2: Open Photoshop with one of your TIFF files, it doesn't matter which file, we won't be saving this one, we just need it to test our settings with, then select Window > Actions.
Step 3:Â From the Actions palette, click on the Create new action button, give it a name, and click on Record. From this point on, everything you do to your document will be recorded into this Action. It's a good idea to explore the settings you want to apply first, before you start recording your action.
Step 4:Â At this point you will need to begin adjusting the Brightness/Contrast and the Exposure settings, you can find them in the Image menu. If you have Live Preview on you can see what works for you. I played with these settings before actually recording my batch action so I'd know which setting worked best for my document.
In my case, adjusting the Exposure to it's maximum setting (+20.00) and the offset to it's lowest setting (-0.5000) seemed to work perfectly, darkening the text, and making the background page brighter.
Step 5:Â When everything looks good with your document just go to the Actions palette and click Stop.
Now that we have our created Action, we can use it to process all the TIFF files and perform the same action on every file. If you are working with a 600 page textbook, your life just got a lot easier.
To run the batch action on your TIFF files, do the following;
Step 1: Click on File > Automate > Batch and select your newly created action under the Play section at the top.
Step 2: Select your Source folder (The one you saved all your TIFF files to.) Make sure you check off the suppress warnings and options. This will prevent the batch action from continually asking you to click on popup boxes.
Step 3: Select your destination folder. I'd recommend creating a new folder to save all the image corrected TIFF files.
All you have to do now is sit back and watch the computer do all the work. Now is a great time to go get a coffee.
When Photoshop has finished working it's magic, all you need to do is go into the destination folder, select all the files (Keyboard Shortcut: Control A) and Right click. From the now open context menu, choose to combine all the files. Adobe Acrobat will now create a single PDF file with all the pages back in order.
This newly created document can now be run through your OCR software of choice and if your Photoshop adjustments worked, should process at a higher success rate.
Your success rate will vary depending on the quality of your original document, but by using these techniques you should hopfully find some kind of improvement.
Here is a sample from my original text.
Here is the corrected text sample.
Hope this helps. If you have any questions, just let me know.