Image vector quantization using neural networks and simulated annealing

Vector quantization (VQ) is a very powerful data compression technique. A number of new approaches to codebook generation methods using neural networks (NN) and simulated annealing (SA) are presented and compared. The authors discuss the competitive learning algorithm (CL) and Kohonen self-organizing feature maps (KSFM). The algorithms are examined using a new training rule and comparisons with the standard rule are included. A new solution to the problem of determining the `closest´ neural unit is also proposed. The second group of methods considered are all based on simulated annealing (SA). A number of improvements to and alternative constructions of the classical `single path´ simulated annealing algorithm are presented to address the problem of suboptimality of VQ codebook generation and provide methods by which solutions closer to the optimum are obtainable for similar computational effort