Performance Evalution of LBG, PSO-LBG, BA-LBG Algorithm

Abstract

Compressing an image refers to reducing the redundancy to efficiently store and transmits data. Vector Quantization (VQ) is a technique based one block coding for compressing image. In this paper, comparison between LBG, PSO-LBG and BA-LBG algorithm has been performed. LBG algorithm is performed in three steps: Vector formation, Codebook generation and Quantization. LBG algorithm can assure that the error will never increment from iteration to next iteration but the algorithm depends largely on the selection of initial codebook. So optimization algorithms were incorporated into LBG algorithm to overcome the dependence on initial codebook. Here, optimization algorithms used are PSO and BA algorithm which is based on the behavior of swarms and echolocation feature of micro bats. The codebook from LBG algorithm is used as the initial codebook for both PSO and BA algorithm. In both algorithms best codebook is chosen based on the fitness value and the process repeats until the stopping criteria is reached. The performance comparison between LBG, PSO-LBG and BA-LBG is done based PSNR and MSE values where BA-LBG algorithm obtains higher PSNR value than LBG and PSO-LBG algorithm.