Abstract
The wavelet-based image compression algorithm has elegant performance and has been widely utilized nowadays. In the paper we propose an image coding algorithm with hybrid wavelet packet transform. For its property of multiresolution analysis and adaptive decomposition on any subband of the whole image, we can use it to preserve the high frequency information of the image than those using wavelet transform such as texture. Besides, the hybrid wavelet packet which is well suited for representation of particular signals allows the choice of multiple appropriate quadrature mirror filters(QMFs) to be included within the analysis instead of the single one. The top-down best basis selection scheme that optimizes hybrid wavelet packet bases makes the representation of image always compact. In order to generate the embedded bit stream for progressive transmission, the transformed coefficients are successive approximation quantized and then coded with context-based arithmetic coding algorithm. Experimental result shows the better compression efficiency than some state-of-the-art coders such as Embedded zerotree wavelet(EZW), Set partitioning in hierarchical trees(SPIHT), and Morphological representation wavelet data (MRWD).