Fully spatial and SNR scalable, SPIHT-based image coding for transmission over heterogenous networks

Abstract

This paper presents a fully scalable image coding scheme based on the set partitioning in hierarchical trees (SPIHT) algorithm. The proposed algorithm, called fully scalable SPIHT (FS-SPIHT), adds the spatial scalability feature to the SPIHT algorithm. It provides this new functionality without sacrificing other important features of the original SPIHT bitstream such as: compression efficiency, full embeddedness and rate scalability. The flexible output bitstream of the FS-SPIHT encoder which consists of a set of embedded parts related to different resolutions and quality levels can be easily adapted (reordered) to given bandwidth and resolution requirements by a simple parser without decoding the bitstream. FS-SPIHT is a very good candidate for image communication over heterogenous networks which requires high degree of scalability from image coding systems.