| 摘要 |
<p>A coding method comprises: organising image data into an array of blocks; transforming data of each block according to a discrete cosine transform, resulting in at least one DC coefficient and at least one AC coefficient for the respective block; for each block X in the image data, determining a first gradient of DC coefficients between a block B above and to the left of block X and a block A immediately to the left of block X; determining a second gradient of DC coefficients between the block B and a block C immediately above block X; when the first gradient is less than the second gradient, predicting a DC coefficient of block X to be the value of the DC coefficient of block C; otherwise, predicting the DC coefficient of block X to be the value of the DC coefficient of block A; and generating a residual DC coefficient for the block X from the predicted DC coefficient and the DC coefficient for block X; and variable length coding coefficients for each block using the residual DC coefficient of each block in place of the DC coefficient for the block, wherein, when any block A, B, C is outside a boundary of the image data, the block's DC coefficient is taken to be a predetermined value. <IMAGE></p> |
