Efficient block entropy coding with low complexity

Efficient low-complexity block entropy coding requires careful exploitation of specific data characteristics to circumvent the practical difficulties associated with large alphabets. Two recent image coding methods, alphabet and group partitioning (AGP) and set partitioning in hierarchical trees (SPIHT) can be viewed as block entropy coding methods, which are successful because of the manner in which they partition the alphabet into sets and encode these sets very efficiently. Here we present analysis and numerical results to show that AGP and SPIHT are indeed efficient block entropy coders.