Modularised fast polynomial transform algorithms for two-dimensional digital signal processing

Novel, modularised, fast polynomial transform (FPT) algorithms for computing 2-D convolutions and 2-D discrete Fourier transforms are presented. In these new methods, only fast polynomial transforms, fast Fourier transforms and number theoretic transforms of the same length are required. Consequently the modularity and regularity of the proposed algorithms make them of great practical value in the concurrent implementation of the parallel-pipeline FPT for 2-D digital signal-processing tasks.