OCR Accuracy

OCR accuracy can be very good. Follow these guidelines to keep accuracy as high as possible.

Handwriting

OCR can only recognize typewritten text, not handwriting.

Page Orientation

OCR requires that the page be upright. If your scanner always produces rotated pages, you can set your conversion job to turn them upright.

Document Quality

One of the most important factors in OCR accuracy is the quality of the original document. Small fonts and rough, blurred characters will degrade OCR accuracy, as will stray marks and excessive "noise" or background patterns. The best accuracy will come from a clean document which uses a 10-14 point font size. The common font faces – Times, Arial, and Courier – will have the highest recognition rates.

Image Resolution

For best accuracy, we recommend that you scan in black and white at 300 dpi. Higher resolution will not greatly improve accuracy but it will increase the file size dramatically. It is not necessary to scan in color or gray scale, which both produce very large file sizes.

Black/White Threshold; Darkness

A very common cause of bad OCR accuracy is an image that is too light. When an image is too light, there can be gaps and holes in the characters. These gaps and holes make it very difficult for the OCR engine to figure out what the characters are. Increase the black/white threshold or darkness setting of your scanner until it produces an image with solid characters.

Conversely, an image that is too dark has very thick characters which bleed together, making it hard for the OCR engine to distinguish one character from the next. Decrease the black/white threshold or darkness setting of your scanner until it produces an image with solid but distinct characters.

OCR Engine

Your choice of OCR engine can make a difference both in OCR accuracy and speed. See Job Advanced Options for more information.