A new class of interpolating wavelets
Jonathan Quevedo, Ioannis A. Kakadiaris, Zhenghao Shi, Donald Jack Kouri · 2003
We present a method for producing a new class of nonseparable interpolating wavelets. The method is based in the property that the scaling function behaves as a generalized delta sequence which allows the construction of the father wavelet using a novel polynomial functional-distributed approximating functional (DAF) and the Quincunx sampling. We present encouraging results from the application of these wavelet-DAFs to the problem of MR image interpolation. The algorithm exhibits an average PSNR improvement of 3dB over bilinear and bicubic techniques.