Splines Computation by Subdivision

In this chapter, fast stable algorithms are presented, which compute splines’ values at dyadic and triadic rational points starting from their samples at integer grid points. The algorithms are implemented by the causal-anticausal recursive filtering of initial data samples, which is followed by iterated application of FIR filters. Extension of the algorithms to the multidimensional case is straightforward. A natural application of the presented subdivision algorithms is for upsampling of signals and images. A few upsampling examples are provided.