Set partitioning in hierarchical trees

From Wikipedia, the free encyclopedia

Set Partitioning in Hierarchical Trees (SPIHT) is an image compression algorithm that exploits the inherent similarities across subbands in a wavelet decomposition of an image. It implies uniform quantization and bit allocation applied after wavelet decomposition.

[edit] General description

The algorithm codes the most important (in the sense of MSE reduction) wavelet transform coefficients in priority, and transmits the bits so that an increasingly refined copy of the original image is obtained with time.

The SPIHT is considered the premier state-of-the-art algorithm in image compression, and has excellent coding performance for 1-D signals. This algorithm has been modified gradually; some of its modified versions: ESPIHT[1], MSPIHT and etc.

The order in which coefficients are transmitted is recovered on the decoder using information of comparisons and sets being examined for significance during the sort, sets are created using hierarchical tree structure, i.e. Set Partition in Hierarchical Trees.

One of the advantages with SPIHT is that it produces an (optimal) embedded bitstream. This means that the bitstream can be truncated at any instant, and is then guaranteed to yield the best possible reconstruction.