Abstract:

This paper proposes a shape-adaptive embedded wavelet image-coding algorithm to code arbitrarilyshaped region. The algorithm introduces the significance state matrixes of coefficient and offspring aggregate according to the LMZC （ Low Memory Zerotree Coding） algorithm to reduce the memory requirement, and applies the maximum amplitude variable to avoid the repeating scan of offspring coefficients. In the coding process, if the coding spending of tree structure is greater than the plus, the bit plane method is adopted to code the wavelet coefficients. Experimental results show that the proposed coding algorithm possesses high implementation speed with small working memory, and that, as compared with LMZC and SPIHT （Set Partitioning in Hierarchical Trees） algorithms at the same comp Signal to Noise Ratio）, th ges with compression ression ratio, it may result in better image reconstruction quality and larger PSNR （Peak us coming to the conclusion that the proposed algorithm can code arbitrarily-shaped imaefficiency.

Key words: image coding, arbitrarily-shaped region, shape-adaptive discrete wavelet transform, zerotree