MODELING A 2-D INVERSE FAST COSINE TRANSFORM ALGORITHM ON A MULTISTAGE NETWORK

The 2-D fast cosine transform is increasingly used in data compression, image coding, signal filtering and feature extraction. In order to implement the transform in parallel processing environment, the signal flow graph of 2-D IFCT has been modelled. Among several interconnection networks for parallel processing, the Omega network is popular due to its property of universality. This paper models the 2-D IFCT algorithm to map on multiprocessors connected through Omega multistage interconnection network (MIN). The communication complexities involved in this mapping have been estimated. The speedup and efficiency performance of Omega network based multiprocessors have been evaluated.