A new set partitioning method for wavelet-based image coding

The "set partitioning in hierarchical trees" (SPIHT) algorithm for wavelet-based image coding, suggested by Said and Pearlman (1996), achieves excellent rate-distortion efficiency while retaining an attractive embedded code property useful for progressive transmission. In the present work we investigate several alternative set partitioning coding strategies in an effort to improve upon SPIHT. The general thrust of our research aims at increasing the size of the sets for which coefficient "significance information" is efficiently transmitted. While we suggest several novel coding ideas for achieving this objective, the resulting method is not found to provide a performance advantage over SPIHT. Some explanation for this observed performance is then provided.