Lossy Lempel-Ziv on subband coding of images

The use of a lossy Lempel-Ziv (LLZ) algorithm as the data compression stratagem on image subband coders (SBC) is discussed. Two LLZ schemes, namely LLZ/s and LLZ/d are proposed. The idea is to replace in the usual SBC, the bank of quantizers followed by Huffman encoder which are used to compress the image components at the output of the several quadrature mirror filters (QMF) or, alternatively, to replace the vector quantizer, by an LLZ. Design complexities inherent to the SBC/QH such as statistical modeling of the filters outputs and bit allocations or the vector quantizer design can be superseded by the SBC/LLZ scheme which dispenses with the knowledge of statistics of the subband being compressed. Simulation analysis of these preliminaries SBC/LLZ schemes have shown that an adequate performance, as compared to other schemes, can be obtained. The two LLZ versions proposed are described and some image compression performance rates are examined.<>