Pruned DCT Algorithm, Digital Signal Processing 6 other important aspect of DCT is its ability to quan- (1996), 145-154. tize the DCT coefficients using visually weighted quantization values. A new algorithm that computes the 8 1 8 pruned Discrete The two-dimensional DCT can be computed using Cosi
A Parallel Algorithm for 4×4 DCT
✍ Scribed by J. Jiang
- Publisher
- Elsevier Science
- Year
- 1999
- Tongue
- English
- Weight
- 175 KB
- Volume
- 57
- Category
- Article
- ISSN
- 0743-7315
No coin nor oath required. For personal study only.
✦ Synopsis
By developing a generalized 1D approach and parallel computing algorithm, this paper presents a parallel algorithm design and hardware implementation for the computation of 4_4 DCT. This algorithm sorts all the 2D input pixel data into four groups. Each group is then forwarded to a 1D DCT arithmetic unit to complete the computation. After a few simple additions which are designed to follow the output of 1D DCTs, the computation of 2D DCT is implemented in parallel. Therefore, the efficiency of the algorithm is entirely dependent on the 1D DCT algorithm adopted, and all the existing fast algorithms for 1D DCT can be directly applied to further optimise the algorithm design. The development can also be further extended to compute general 2D DCT by a recursive procedure where the 4_4 DCT algorithm is used as the basic core.
📜 SIMILAR VOLUMES
## Abstract An algorithm, which utilizes a high degree of parallel processing, has been developed for two‐dimensional rotating‐frame zeugmatography so that a picture of 256 X 256 pixels can be generated with a minicomputer system 2 sec after data accumulation. An array processor is employed as a se
We present a parallel randomized algorithm running on a CRCW PRAM, to determine whether two planar graphs are isomorphic, and if so to find the isomorphism. We assume that we have a tree of separators for each planar graph Ž Ž 2 . 1 q ⑀ which can be computed by known algorithms in O log n time with
In this paper we propose a new parallelization of the Davidson algorithm adapted for many eigenvalues. In our parallelization we use a relationship between two consecutive subspaces which allows us to calculate eigenvalues in the subspace through an arrowhead matrix. Theoretical timing estimates for