Zero and first-degree spline multiwavelets and parallel computing

Abstract

Abstract The paper considers the Haar-like multiwavelets of zero degree splines (step functions) and first-degree splines (broken lines). In contrast to the use of the wavelet transforms based on the Hermitian spline-multiwavelets, the approach proposed does not require the calculation of approximate values of derivatives. On the other hand, the parallelization effect does not rigidly relate to the degree of a spline. This can lead to a significant computation speedup due to the parallel processing of the measured values by several (instead of one) filters. The results of numerical calculations are compared to the known analogs.