Efficient filtering structure for spline interpolation and decimation
An efficient structure for spline-based fractional delay filtering for interpolation/decimation is introduced. Inspired by the Newton structures for Lagrange interpolation, it requires less than half the number of operations of a typical Farrow implementation. Moreover, it displays better frequency response characteristics than Lagrange-based filters. To obtain this structure, a matrix form of the Farrow transfer function is proposed and used to derive state-space transformations between the Lagrange-Farrow structure and its Newton counterpart. These transformations are then applied to the spline polynomial, giving rise to the efficient Newton-like spline filtering method.