| 摘要 |
PURPOSE:To improve picture quality by vector-quantizing an orthogonally transformed coefficient to a representative vector number in accordance with the character. CONSTITUTION:A vector quantizing device 108 vector-quantizes the coefficient from an orthogonal transform circuit 406 to the representative vector number, and outputs it to a line and a vector quantizing decoding circuit 110. The representative vector is decided in accordance with the statistical property of the orthogonally transformed coefficient. The vector quantizing decoding device 110 reproduces the coefficient into the orthogonally transformed coefficient by the representative vector number from the vector quantizing device 108. An orthogonal inverse transform circuit 412 reproduces the decoded coefficient into a predictive error. An adder 414 adds the predictive error reproduced by the orthogonal inverse transform circuit 412 and that from a predicting circuit 419, reproduces a picture element value and writes it into a frame memory 417. Thus, the deterioration of picture quality is prevented to a part where the predictive error signal has much low frequency component such as the flat part of a picture, and a part where the predictive error signal has much high frequency component such as the edge part of the picture. |
