Object-based VQ for image compression

In this paper, a vector quantization algorithm is proposed for image compression. The proposed algorithm composed of three phases, initialization, iterative and finalization. The initialization phase based on Max–Min algorithm. The iterative phase is an adaptive LBG algorithm. The finalization frees the codebook from redundancy. The LBG adapted to locate the codebook points to line up across the boundary of the objects they represent. The algorithm testing results showed that images main features sustain the high compression ratios. The same algorithm is applicable to codebook bins to improve the quality for the areas that have fine details.