Efficient representation and decoding of static Huffman code tables in a very low bit rate environment

The lossless entropy coding used in many image coding schemes often is overlooked as most research is based around the lossy stages of image compression. This paper examines the relative merits of using static Huffman coding with a compact optimal table verses more sophisticated adaptive arithmetic methods. For very low bit rate image compression, the computationally simple Huffman method is shown to be competitive and often superior to adaptive algorithms. We present a method of efficiently representing an optimal Huffman table using delta coded symbol bit lengths. The decoding of the bitstream can also be accelerated by using table lookup operations.