Automatic Generation Of OCR Logic From Scanned Characters
Original Publication Date: 1979-Aug-01
Included in the Prior Art Database: 2005-Feb-20
This article describes a method for adaptively modifying the sequential optical character recognition (OCR) algorithm for an OCR device as the type font of the characters to be recognized changes. The method step involves the estimation of picture element (PEL) statistics for each character type and the alteration of the OCR device decision tree utilizing the statistics. A system implementing this method could consist of an OCR unit employing conventional fonts and apparatus for scanning, thresholding, segmenting, and registering printed characters. The OCR unit should classify individual character patterns by way of a decision tree stored in a read/write memory. Such a tree is shown, for example, in Fig. 1.