Character Images Compression by Adaptive Arithmetic Coding and its Application to the TV Prompt Reporting of the Election of the House of Councilors in 1998

Abstract

A new method of character image compression, for use in a system for the prompt reporting of election results on broadcast TV, has been developed. It compresses character images without information loss using adaptive arithmetic coding and 2-D spatial resolution conversion. The encoder draws kanji characters onto an oversized bitmap domain and compresses this using adaptive arithmetic coding. The decoder converts the image from binary to grayscale at a reduced spatial resolution appropriate for TV broadcasting. The method has been adopted for KINDS, a system for the prompt reporting of election results on NHK TV, and was applied to the election of the House of Councillors on 12 July 1998. The characteristics of the method are discussed in relation to the context length of the coding and the nonstationarity of the images using the actual results of the election.