A Very Low Bit Rate Image Compressor Using Transformed Classified Vector Quantization

In this paper, a very low bit rate image compression scheme is proposed. This scheme is a hybrid method that combines the good energy-compaction property of DCT with the high compression ratio of VQ-based coding. We start by transforming image block from spatial domain to frequency domain using DCT. In order to increase the compression ratio while preserving decent reconstructed image quality, DC coefficients are coded by DPCM and only the most important AC coefficients are coded using classified vector quantization (CVQ). The most important AC coefficients are selected to train the codebook according to the energy packing region of different block classes. Also, this scheme can provide different compression ratios like JPEG does with the same codebooks. The experimental results show that the proposed scheme performs much better than JPEG at low bit rate.