Codebook Design Optimization Based on Estimation of Distribution Algorithms

Vector quantization has been a very important technique for compressing both the image and the speech data. One of the key problems arising in vector quantization is the codebook design problem. In this paper, use estimation of distribution algorithms (EDAs) to optimize codebook design. EDAs are evolutionary computation, combined by genetic algorithm and statistically learning. In order to verify the EDAs performance, compare the EDAs with LBG and GA. The experiment results show that the EDAs make better performance on improving the codebook quality.