| 摘要 |
PURPOSE:To prevent a quantization error from being increased excessively by transmitting an input vector when it is larger than a distortion threshold value in which the minimum distortion is set, and storing it by rewriting to a quantization vector used most frequently in the past in a code book as a new quantization representative vector. CONSTITUTION:The input vector 6 extracted based on the minimum distortion at the time of performing vector quantization is transmitted, and also, the input vector 6 is stored by rewriting to the quantization representative vector 7 transmitted most frequently in the past in the code book 2 as the new quantization representative vector 7. Then, the code book 2 is prepared as an updatable random access code book. In such a way, it is possible to obtain a vector quantizer in which the quantization error for a specific input vector can sufficiently be reduced. |
