摘要 |
In a method for decoding images, a previously decoded image is stored as a reference image (210), motion vector information (202) is extracted from input information to be decoded, and motion compensation is performed using the motion vector information and the reference image for synthesising a prediction image (212). Pixel value interpolation for obtaining the prediction image is performed by using a rounding method specified by rounding method information, which is included in a header section of input information of the currently decoded image and specifies whether positive rounding or negative rounding is used. |