Parallel vector processing of multidimensional orthogonal transforms for digital signal processing applications (Q2277807)

The paper presents vector and parallel algorithms and implementations of one- and two-dimensional orthogonal transforms. The sinusoidal transforms (discrete cosine transform, discrete sine transform, discrete Hartley transform) are computed via fast real discrete Fourier transforms. The discrete Walsh transform and the discrete Hadamard transform are performed by using direct factorizations of transform matrices. The two- dimensional transform algorithms are based on the row-column method. The speed performances are evaluated on Cray X-MP/48 vector computer.