| 摘要 |
<p>PURPOSE: To provide an encoded image having objectively superior quality by performing information allocation and encoding control for the unit of a macro block corresponding to a DCT AC coefficient reflected with the local definition degree of image. CONSTITUTION: An input image 1 is transformed for each small block by a DCT part 2. On the other hand, a predictive image 4 between motion compensation frames is used for the inter-frame prediction of input image 1, and an inter-frame differential image 5 is generated. This differential image 5 is inputted to a DCT part 6 for each block and a DCT coefficient 7 of differential image is calculated. A DCT coefficient 3 of input image and the DCT coefficient 7 of differential image are inputted to a variance computing part 8, and variance 9 of DCT AC coefficient is calculated for the unit of a macro block. Then, concerning the variance 9, a quantizing step 35 is found by an encoding control part 10. At a quantizing part 11, the DCT coefficient 7 is quantized while using the quantizing step determined by the encoding control part 10, and variable length encoding is performed. Thus, the quality of encoded image can be improved.</p> |
