Going Gray? Comparing the OCR Accuracy Levels of Bitonal and Grayscale Images by Tracy Powell and Gordon Paynter. Both from the National Library of New Zealand.
The authors processed the same items in bitonal (black and white) and grayscale, then OCRed them. They then counted up the OCR errors. The theory was that Grayscale images would give better results because it gave the OCR software more information to work off of. They were surprised by the outcome. From their experiment, it seems as if there is no difference in OCR quality.
This is interesting to me mostly because I had heard that bitonal images gave better OCR results.For a review of how OCR software works, there is a great explanation on Mustek.com: Understanding OCR.
The authors processed the same items in bitonal (black and white) and grayscale, then OCRed them. They then counted up the OCR errors. The theory was that Grayscale images would give better results because it gave the OCR software more information to work off of. They were surprised by the outcome. From their experiment, it seems as if there is no difference in OCR quality.
This is interesting to me mostly because I had heard that bitonal images gave better OCR results.For a review of how OCR software works, there is a great explanation on Mustek.com: Understanding OCR.
Comments