Details:
Year: 1995 Vol. 1
Pages: 18
Summary:
The conversion of printed material to digital data has been heavily studied from, particularly, the perspective of pattern recognition and image signal processing. It is the purpose of this research to briefly examine two new approaches to optical character recognition (OCR) production systems: 1) a post scanning, post OCR application that provides a perfect or near perfect conversion of printer or offset printed derived originals to page described, text editable digital data and 2) a neural network based system that provides highly accurate conversion of geometrically complex documents.