B-spline signal processing is an important tool in signal processing society. B-splines can be adapted in interpolation and approximation of underlying continuous signals and systems, adjustment of fractional time delays in communication devices and in numerical signal analysis. In this work, we introduce the shifted B-spline interpolation algorithm, where the B-spline kernel is shifted by a fraction D of the sampling period. Experimental tests show that the interpolation error is strongly dependent on Δ and attains a minimum at Δ = 0.25. We describe the z-transform filter for efficient parallel implementation of the shifted B-spline interpolation algorithm. The performance of the D-shifted B-spine interpolation filter is significantly higher compared with the cubic convolution interpolation and the standard B-spline interpolation.
|Journal||Current Development in Theory and Applications of Wavelets|
|Publication status||Published - 2010|
|MoE publication type||A1 Journal article-refereed|