ABSTRACT
The two-dimensional discrete cosine transform (2D-DCT) of a square or a rectangular block shape is used for almost all block-based transform schemes for image and video coding. The 2D-DCT can be separately implemented through two 1D transforms, one to process the block along the vertical direction and another along the horizontal direction. When the conventional discrete cosine transform (DCT) is applied to an image block in which other directional edges dominate, the non-zero coefficients are not well aligned across different columns or rows and the 1D DCT may produce more non-zero coefficients. The direction of transform is chosen according to the dominating edges. Domminant edges within each individual image block. Coefficients produced by all directional transforms are arranged appropriately; a second transformation can be applied to the coefficients that are best aligned with one another. The steerable DCT When block-based separable transforms are used, the resulting coefficients tend to represent inefficiently the block when it contains arbitrarily shaped discontinuities.
