Memory efficient image coding with embedded zero block-tree coder
H. Arora, Pushpendra Singh, Ekram Khan, Fay Ghani · 2005
The work presents an embedded and memory efficient image compression algorithm which exploits both inter- and intra-band correlation of wavelet coefficients. Set partitioning in hierarchical tree (SPIHT) is a zero-tree based coder which exploits inter-band correlation among bands of the same orientation, while the set-partitioning embedded block coder (SPECK) is a zero-block based coder which exploits intra-band correlation. However, they have extensively large memory requirements due to the use of three/two linked lists whose entries increase from one-bit-plane to the next. We propose an algorithm that is based on block-set partitioning and quad-splitting using two re-usable lists. The main list is initialized at the beginning of each bit-plane and is exhausted within the same bit-plane. This makes our proposed algorithm highly memory efficient. Experimental results show that the compression efficiency of the proposed method is comparable to any state-of-the-art image coder while reducing the memory requirement by 50-60% in comparison to the SPIHT algorithm.