Finite state hierarchical table-lookup vector quantization for images

This paper presents an algorithm for image compression using finite state hierarchical table-lookup vector quantization. Finite state vector quantizers are vector quantizers with memory. Finite state vector quantization (FSVQ) takes advantage of the correlation between adjacent blocks of pixels in an image and also helps in overcoming the complexity problem of block memoryless VQ for large block sizes by using smaller block sizes for similar performance. FSVQ algorithms typically try to preserve edge and gray scale gradient continuity across block boundaries in images in order to reduce blockiness. Our algorithm combines FSVQ with hierarchical table-lookup vector quantization. Thus the full-search encoder in an FSVQ is replaced by a table-lookup encoder. In these table lookup encoders, input vectors to the encoder are used directly as addresses in code tables to choose the code-words. In order to preserve manageable table sizes for large dimension VQs, we use hierarchical structures to quantize the vector successively in stages. Since both the encoder and decoder are implemented by table lookups, there are no arithmetic computations required in the final system implementation. To further improve the subjective quality of compressed images we use block transform based finite-state table-lookup vector quantizers with subjective distortion measures. There is no need to perform the forward or reverse transforms as they are implemented in the tables