| 摘要 |
<p>The data compression ratio is increased while the high quality of image of image data is maintained, and the calculation load in decoding is lightened. In Figure 1(A), image data is divided into blocks, a DC image is formed of t he average values of the blocks, the block average values are separated from pixel blocks, and a residual vector for each block is found. If the magnitud e of the residual vector is the allowable value or more, an orthogonal base fo r approximation of the residual vector is found by adaptive orthogonal conversion using a nest of the DC image and coded. In Figure 1(B), if the residual vector is less than the allowable value, pixel data on 4x4 pixels o f a block S is directly reproduced on the decoder side, instead of coding the orthogonal base, by a nonstep alternating current component predicting metho d based on five sets of DC image data on the block S and upper, lower, left, a nd right blocks (U, B, L, R) formed of block average values of 4x4 pixels.</SDO AB></p> |
