A new SPIHT algorithm based on variable sorting thresholds

The "set partitioning in hierarchical trees" (SPIHT) algorithm for wavelet-based image coding, introduced by Said and Pearlman (1996), achieves excellent rate-distortion efficiency and offers an attractive embedded data property to allow fully scalable transmission. In this paper, we present a more efficient implementation of the SPIHT algorithm by using variable thresholds to sort the list of insignificant pixels (LIP) and the list of insignificant sets (LIS). In our work, we observe two interesting facts: (1) most of the initial subsets in LIS are not only insignificant with respect to the maximum threshold, but also insignificant with respect to a smaller threshold; and (2) most of the pixels generated from sorting LIS are smaller than the current threshold. Based on these two observations, we propose a new image codec, which can make the binary uncoded outputs more efficient, and can work well on different image sizes and different decomposition levels.