On Generalizations and Improvements to the Shannon-Fano Code

Abstract

This paper examines the possibility of generalizing the Shannon-Fano code for cases where the output alphabet has more then 2 (n) symbols. This generalization is well-known for the famous Huffman code. Furthermore, we will be looking at possible improvements to the algorithm, as well as other entropy based lossless data compression techniques, based on the same ideas as the Shannon-Fano code. All algorithms discussed in the paper were implemented by us in C++, and we will be illustrating our hypotheses with test results, made on an average performance PC.