| 摘要 |
<p>PURPOSE:To prevent the deterioration of a picture with respect to multi-level data and to enhance the efficiency of coding by separating plural components from data which have been pre-processed and severally coding the components. CONSTITUTION:Multi-level picture information is segmented into blocks consisting of picture elements of (m)X(n) and the blocks are assumed to be matrixes. An orthogonal transformer 1 executes Hadamard-transformation and a scalar quantizer 2 executes a pre-processing for scalar qantizaiton. A separator 3 separates plural components from respective elements of the matrixes pre-processed into negative positive code (phase) information P, absolute value information A and information of zero or non-zero (zero information Z). The separated information are transmitted to vector quantizers 4-6 at every block and are respectively vector-quantized. Since vector-quantization is executed in addition to scalar quantization in units of blocks of (m)X(n), coding is efficiently attained.</p> |
