Self-organized dynamic Huffman coding without frequency counts

Abstract

Summary form only given. Dynamic Huffman coding uses a binary code tree data structure to encode the relative frequency counts of the symbols being coded. The authors aim is to obtain a simple and practical statistical algorithm in order to improve the processing speed while maintaining a high compression ratio. The algorithm proposed uses a self-organizing rule (transpose heuristic) to reconstruct the code tree. It renews the code tree by only switching the ordered positions of corresponding symbols. This method is called self organized dynamic Huffman coding. To achieve a higher compression ratio they employ context modelling.