A Strategy of Particle-Pair for Vector Quantization in Image Coding

This paper presents a new strategy of particle-pair(PP) for vector quantization(VQ) in image coding.In this strategy,two particles are combined into a particle-pair based on conventional particle swarm optimization(PSO) algorithm.At each iteration,the particle-pair performs basic operations of PSO(velocity updating and position updating) and conventional LBG algorithm in sequence.The codevectors flying over the boundary are replaced with the training vectors,which have large distortions.This strategy prevents the particle from being trapped in a local optimum,memorizes and estimates the best direction the particle moves toward to find the optimum codebook design.The codevectors are scattered reasonably both in high density distribution regions and low density areas of the training vector space.Experimental results have demonstrated that the performance of this new algorithm is much better than that of FKM,FRLVQ,FRLVQ-FVQ consistently with shorter computational time and higher convergence rate,and the dependence of the final optimum codebook on the selection of the initial codebook is reduced effectively.