Extract from "A Conclusive Methodology for Rating OCR Performance"

Introduction
Efficient, reliable, and fully automated optical character
recognition (OCR) has become one of the most important
problems in modern document analysis. OCR is a method of
transforming a page image into a text file. The goal of this
transformation is to make letters, words, and symbols
printed on a page identifiable. Document rating systems
attempt to rank page images on the basis of the degree to
which they can be accurately transformed into text by OCR.

Nathan E. Brener, S.S. Iyengar, and O.S. Pianykh
Department of Computer Science, Louisiana State University, Baton Rouge, LA 70803.
E-mail: brener@bit.csc.lsu.edu