摘要 |
PURPOSE:To improve the coding efficiency by taking a local property of an input vector into account so as to vary the processing block size and using plural code books so as to apply vector quantization processing. CONSTITUTION:A picture input signal by one frame is stored in a 1st frame memory in an input frame memory 1. A block control circuit 2 arranges plural picture signals each, applies block processing in a variable size to generate an input vector 4. Then the input vector 4, its address 5 and block size information 3 are outputted. A vector quantization coding circuit 6 applies vector quantization coding to the input vector based on any information of 1st-3rd code books 14-10 and information from the circuit 2. When a minimum distortion at quantization is larger than the set threshold level, the input vector area is use as a specific block. It is used as a representative vector 10 to give a code book number 7, an address 8 and an index 9 to a memory 13 of a 2nd frame memory and a code book control circuit 12 to generate 2nd and 3rd code books 15, 16. |