This paper discusses the recursive implementation of the discrete cosine transform (DCT) and its inverse (IDCT). The transform is constructed by using recursive filter structure to generate the transform kernel values. We first derive two trigonometric equations, which can be represented as the Chebyshev polynomial. Then we demonstrate that general length of the DCT and IDCT can be efficiently implemented by using the regressive structure derived from the recursive formulae. The computational complexity of each data throughput in these architectures is less than that in the conventional ones by as many as 50%. The proposed architectures are regular and suitable for parallel VLSI implementation.
|Number of pages||11|
|Journal||IEEE Workshop on Signal Processing Systems, SiPS: Design and Implementation|
|Publication status||Published - 1999 Dec 1|
|Event||1999 IEEE Workshop on SiGNAL Processing Systems (SiPS 99): 'Design and Implementation' - Taipei, Taiwan|
Duration: 1999 Oct 20 → 1999 Oct 22
All Science Journal Classification (ASJC) codes