摘要 |
<p>PURPOSE:To increase the formation efficiency and versatility of the set of an output vector by forming at high speed the output vector exclusive for every picture, and coding the vectors obtained by including this vector and the results of the quantified and coded vector and outputting them. CONSTITUTION:Picture signal series 15 are stored in a picture memory 16, are converted for every K pieces, to blocks, and are read as an input vector 1. A vector quantification part 17 outputs an index 18 of the output vector which gives the minimum distortion to the input vector 1. A code table renewal part inputs the input vector 1 and the vector quantification index 18. The input vector which is read from the picture memory 16 is again quantified in terms of vector by using the set of a new output vector. When clustering is executed by specified times, a latch 21 outputs an output vector set 20 to the transmission route, and outputs the vector quantification index 18 of the picture which uses the set.</p> |