Improvement and Comparisons of Several Soft Computing Approaches for Vector Quantization Codebook Generation

In this paper, we review several soft computing methods and GLA for codebook generation problem of vector quantization (VQ) and improve some of them. The codebook generation algorithms discussed includes Generalized Lloyd Algorithm (GLA), Kohonen Self Organized Feature Map (SOM), Fuzzy c-means clustering, Genetic Algorithm (GA) and Hierarchical SOM (HSOM). This paper improves genetic algorithm with a suitable crossover and SOM using four techniques.