Improved dynamic bit allocation in image coding using a self-organizing map with learning vector quantization

摘要

A self-organizing map (SOM) neural network for dynamic bit allocation in adaptive image transform coding is presented. The results shown in this paper are for nets with 30, 96 and 128 neurons in the input layer and 100 (10/spl times/10) neurons in the competition layer. The learning vector quantization LVQ1 algorithm was used to enhance the clustering in the map.