Context quantization based on the modified genetic algorithm with K-means

Abstract

In this paper, the context quantization for I-ary sources based on a modified genetic algorithm is presented. In this algorithm, the optimal context quantizer is described by the chromosome which contains the optimal number of classes and the corresponding cluster centers. The adaptive code length is used to evaluate the fitness value to find the best chromosome. The rules for the selection, the crossover and the mutation operations are discussed. A K-means operator is incorporated in each iteration to accelerate the convergence of the algorithm. The optimized context quantizer can be obtained without the prior knowledge of the number of classes. Simulations indicate that the proposed algorithm produces results that approximate the best result obtained by the K-means-based context quantization with lower computational complexity.