| 摘要 |
A discrete cosine transform (DCT) coefficient prediction method and a DCT coefficient encoding method are provided. For the prediction of AC coefficients, location information of the nonzero AC coefficients is used. Then, the AC coefficients are divided into the vertical and horizontal component areas, and a remaining area, and variable modified zigzag scanning is applied for encoding. The performance of the AC coefficient prediction coding is improved as the absolute values of the predicted coefficients are small, and as the number of coefficients of value zero increases. The coefficients of value zero may be predicted as nonzero, so these coefficients are eliminated from the prediction. On the other hand, when using the characteristics of the adjacent blocks in a spatial domain, a larger weighted value is applied to the pixel values which are the nearest to the current block, thereby improving the prediction characteristics for DC/AC coefficients in a discrete cosine transform (DCT) domain.
|
