FAST STRUCTURAL 2-DIMENSIONAL DISCRETE COSINE TRANSFORM ALGORITHMS

The matrix decomposition of transformation associated with the Kroneck er product not only provides a thoughtful structure in hardware realiz ation but also bestows a skillful tool for complexity evaluation. Henc e, there are several fast algorithms developed to achieve efficient co mputation of two dimensional (2-D) discrete cosine transform (DCT) wit h matrix decomposition techniques. However, we found that their deriva tions associated with their computation structures were not shown form ally. In this paper, we propose formal derivations to remedy their def iciencies to achieve more structural 2-D DCT and inverse DCT (IDCT) al gorithms. Furthermore, we also show that the remedied algorithms are w ith less computational complexity and more regular structure for reali zation.