Adaptive image coding based on vector quantization using SOFM-NN algorithm

Vector quantization (VQ) has been applied widely in image coding. The codebook of the VQ system can be generated by a clustering algorithms such as K-Means and LBG algorithms or a feature mapping algorithm such as SOFM-NN. In this paper we present a performance comparison between the LBG and SOFM algorithms in term of PSNR for several bit rates and propose a simple adaptive mechanism for the SOFM codebook. The result leads to a conclusion that LBG and SOFM are comparable, but SOFM has greater advantage in providing simple adaptation mechanisms