Statistical encoding for text and picture communication

Abstract

Suggestions have been made from time to time for reducing the transmission time or bandwidth required for the facsimile transmission of text or pictorial material by representing it by a description which has low redundancy. This paper presents an investigation of the channel capacity required for various types of copy, using one such description, based on one-dimensional differential coordinates together with statistically matched encoding. A binary digital transmission channel is assumed throughout. Only black and white copy is considered. A written or printed message is assumed to consist of a 2-dimensional array of resolution elements, called dots, whose number determines the definition. Since a definition of 0.01 inch is widely accepted as satisfactory for facsimile reproduction of black and white copy, this figure, corresponding to almost a million dots for an 8 1/2- by 11-inch typewritersize sheet, is used throughout.