Interframe coding of images using lattice vector quantization

Vector Quantization (VQ) is a well-established technique for low bitrate source coding due to its inherent theoretical superiority over the scalar quantization. In product code VQ the input vector is decomposed to different elements and separate codebooks are designed for each element. In the paper, a new variation of SVQ is presented, which vastly reduces the encoding complexity. Fast Lattice-based Spherical Vector Quantization (FLSVQ), enables the use of very fast lattice quantizing algorithms instead of the exhaustive search of all the codebook points. Fast lattice nearest neighbour algorithms have been developed by conway and Sloane (1982). In this method an extension of these algorithms is used. Then, a novel two-stage lattice vector quantization method based on FLSVQ is tested for interframe coding of images sequences and simulation results are provided to assess the performance of the proposed method