Setting priorities: a new SPIHT-compatible algorithm for image compression

Abstract

We introduce a new algorithm for progressive or multiresolution image compression. The algorithm improves on the Set Partitioning in Hierarchical Trees (SPIlT) algorithm by replacing the SPIlT encoder. The new encoder optimizes the multiresolution code performance relative to a user-defined probability distribution (or priority function) over the code's rates or resolutions. The new algorithm's decoder is identical to the SPIlT decoder. The resulting code achieves the optimal expected performance across resolutions subject to the constraints imposed by the use of the SPIHT decoder and the distribution (or priorities) over resolutions set by the user. The encoder optimization yields performance improvements at the rates or resolutions of greatest importance (according to the encoder's priority function) at the expense of performance degradation at low priority rates or resolutions. The algorithm is fully compatible at the decoder with the original SPIlT algorithm. In particular, the decoder requires no knowledge of the priority function employed at the encoder. Experimental results on an image containing both text and photographic material yield up to 0.86 dB performance improvement over SPIHT at the resolution of highest priority.

Multiresolution Vector Quantization

“…Using this approach, a dimension-fixed-or variablerate vector quantizer is optimal if it lies on the lower convex hull of or , respectively. This optimality criterion, first introduced in [19], can also be applied in other multiresolution codes [20], [21]. It differs critically from its predecessors, as described in Section III.…”