Particle Swarm Optimization Applied to Image Vector Quantization

Abstract

Codebook design of VQ (Vector Quantization) is a global optimization problem. The LBG algorithm depends upon the initial codebook and is prone to converge to a local optimal solution. To solve the problem, adopt PSO (Particle Swarm Optimization) to design the optimal codebook of image vector quantization and present PSO-VQ (PSO Vector Quantization) algorithm. According to PSO-VQ, a particle indicates a codebook and the optimal codebook is obtained from iterations of the initial codebooks by method of the particle evolvement. To ensure the solution converge to the global optimal codebook, the authors presented the PCO (Particle Coherent Operation), by which the code vectors of each initial codebook are sorted in ascending order based on the average gray value of the pixels in the code vector, and so that the inner structures of all the particles are essentially identical. The experimental results show that the PSO-VQ algorithm is feasible and effective, as well as develops the application of the PSO.KeywordsParticle Swarm OptimizationParticle SwarmVector QuantizationGlobal Optimization ProblemOptimal Power FlowThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.