A wavelet packet image coding algorithm based on SPECK

This paper introduces a new wavelet packet image compression algorithm called WP-SPECK which is based on the set partitioned embedded block coder (SPECK). The zerotree quantization coding which is successful for dyadic wavelet decomposition is not best for WP, because the parent-child relationships of tree structures are difficult to define and the probability of zerotrees are greatly reduced in the WP. To avoid the shortcomings of zerotree quantization, the WP-SPECK algorithm using block quantization shows lower complexity and more efficient performance in wavelet packet image coding. Experimental results demonstrate that the WP-SPECK significantly outperforms the base-line SPECK for PSNR in images of rich texture, and provides better visual quality in texture regions of images.