Embedded Layered Clusters Wavelet Zerotree Image Coding
Zhang Zong · Chinese Journal of Computers · 2002
Theoretical analysis shows that a down-layer cluster always wraps an upper-layer one that having same seed when its structure element no less than one on upper layer. Therefore, by selecting a group of structure elements on successive bit plane layer, a progressive morphological representation for significant coefficients is allowed. Based on this observation, a layered cluster morphological representation is extended to the wavelet zerotree algorithm, namely the embedded layered cluster wavelet zerotree (ELCWZT) coding algorithm. The ELCWZT algorithm exploits not only the zerotree to improve the coding efficiency of the insignificant coefficients, but also the morphological conditioned dilation operator to improve the coding efficiency of the intra-subband significant coefficients. Therefore it can be expected to overcome effectively the shortcomings that exist when using only zerotree data structure or morphological cluster representation. Experimental results on the standard test images show that the proposed ELCWZT algorithm outperforms EZW codec. It provides the performance comparable to that of SPIHT codec as well as MRWD codec in PSNR. The proposed algorithm also has many desirable features, such as the full embeddedness for progressive transmission and CBR traffic, and low-complexity. In addition to the rate scalability, its modified version also allows multiresolution scalability in encoding and decoding.