PROGRESSIVE VECTOR QUANTIZATION ON A MASSIVELY-PARALLEL SIMD MACHINE WITH APPLICATION TO MULTISPECTRAL IMAGE DATA

This correspondence discusses a progressive vector quantization (VQ) c ompression approach, which decomposes image data into a number of leve ls using full-search VQ. The final level is losslessly compressed, ena bling lossless reconstruction. The computational difficulties are addr essed by implementation on a massively parallel SIR ID machine, We dem onstrate progressive VQ on multispectral imagery obtained from the adv anced very high resolution radiometer (AVHRR) and other earth-observat ion image data, and investigate the tradeoffs in selecting the number of decomposition levels and codebook training method.