A New Image Compression Method Based on LBG Algorithm in DCT Domain

Vector Quantization (VQ) is an effective technique for digital image compression. In this paper, we apply a new method based on discrete cosine transform (DCT) and Linde-Buz o-Gray (LBG) algorithm to construct the codebook of vector quantization for image compression. In the proposed method, the image is divided into subimages, then DCT coefficients of subimages are computed, quantized, and truncated. Next, the LBG algorithm is used to generate the codebook in DCT domain and finally, for each subimage a codeword is assigned. The experimental results indicate that the new algorithm achieves high performance from the subjective and objective points of view. Further, the new method outperforms the recently introduced VQ-based methods.