Adjust Contrast Of Scanned Image

  dogbreath1 13:48 PM 25 Nov 12
Locked

I've been doing some local history research which has involved printing out various newspaper articles held on library film. Some of these images are quite dark and scanning the prints prior to OCR'ing them has resulted in black 'text' on a relatively dark grey background. The resultant OCR'd text files are at best full of errors. I'm looking for a way of enhancing the contrast of the scanned images to improve the accuracy of the OCR process. Any ideas??

  wiz-king 14:59 PM 25 Nov 12

What image program are you using?

If you are just using a scanner, then download Irfanview and use that to enhance the image. You may also have trouble with your OCR program not recognising some old typefaces but there's not a lot you can do about that.

  Woolwell 15:03 PM 25 Nov 12

You should be able to scan in black and white only or other settings that may well reduce the problems of black on grey.

  dogbreath1 16:41 PM 25 Nov 12

I have IrfanView, but I cannot work our how to enhance contrast. In an ideal world I would be able to convert the mottled grey background to white!

  wiz-king 17:08 PM 25 Nov 12

Open your image, go to top toolbar >image > color corection and use the contrast and brighness controls > save as same name-1

  dogbreath1 17:20 PM 25 Nov 12

Thanks. That looks promising. I'll give it a try.

  dogbreath1 23:53 PM 25 Nov 12

The IrfanView contrast tweak makes the scanned text look and read much better but actually worsens the OCR results!

Any more ideas, please?

  Simsy 07:52 AM 26 Nov 12

When scanning newspaper text the reverse side of the page often can show through... to avoid this what I always do is have a piece of black card on top of the page that I'm scanning...

This does tend to make the "white" on the page you are scanning darker, but increasing the brightness and contrast of the subsequent image usually produces a good result.

Also, I would scan in "Greyscale" rather than "Black and White", (which can produce a very "blocky" result).

In Irfanview the brightness/contrast adjustments are via the Menu bar under Image>Color Corrections.

Hope this helps.

Regards,

Simsy

  dogbreath1 11:18 AM 26 Nov 12

Thanks for the reply. The newspapers I'm copying are stored on film, so I cannot physically adjust the image result by inserting dark card. I can obtain a print out of these film images but to lighten the background often adversely affects the boldness and readability of the text. Experience to date confirms that greyscale scans of the printouts are superior to black/white, but that colour scans are even better.

  wiz-king 13:00 PM 26 Nov 12

Unfortunately most OCR programs on recognise modern fonts. I have often had this problem, I have a copy of ABBYY finereader 7 that does a fair job of it but it still has its moments.

  dogbreath1 20:38 PM 25 Dec 12

Simsy makes a good point about avoiding scanning in black and white but, as daft as it may sound, I find colour scans of monochrome text images work better than greyscale for the purposes of OCR'ing.

Thank you all for your responses, but it seems there isn't a magic bullet for this problem.

Advertisement

This thread is now locked and can not be replied to.

Sony Xperia Z5 vs Xperia Z5 Premium vs Xperia Z5 Compact: What's the difference between Sony's new ?…

1995-2015: How technology has changed the world in 20 years

This 3D food printer prints chocolate, cookie dough or pizza

iPhone 6S preview: What to expect from Apple's next iPhone