Skip to main content
Kofax

Recommended height (in pixels) of characters for optimal OCR

Recommended height (in pixels) of characters for optimal OCR 

Information:

In this article we provide information about the recommended pixel height of characters for optimal OCR results.

There is a difference between Western and Asian OCR in this aspect, so there are two separate sections below.

Western (Latin, Greek, Cyrillic) characters

You can achieve the best results if the height of the individual characters (e.g. lower case "d") are in the range of 25 to 30 pixels. This requirement is met e.g. when a Letter size page with standard margins is scanned with 300 to 400 DPI, and the page contains Times New Roman characters of 12 point size. In this case there will be 78 characters in a line, and 46 lines in the page.

Asian (Japanese, Chinese, Korean) characters

The ideal font point is 12pt scanned at 300 DPI, approximately 48*48 pixel*pixel.

The ideal DPI is 300 DPI for character font size greater than 10.5pt. It is recommended to use higher resolution (400 DPI or 600 DPI) for character font size less than 10.5pt.

10.5pt is equal to the font size FIVE in Chinese printed document, and 12pt is equal to SMALL FOUR, 14pt for FOUR, .etc. 12pt and 10.5pt are commonly used in text body of Chinese printed document. The phenomenon is similar in Japanese and Korean printed documents.

The acceptable minimal character size is 30*30 pixel*pixel.