Theory and practice of interpolation in the pyramid domain

Algorithm for missing data interpolation

Having a very fine sampling on the $u$ axis attenuates most of the transformation artifacts of the pyramid transform. However, as seen in Figure 3a, some noise due to the linear interpolation operator used for the mapping between the $(\omega ,x)$ and $(\omega ,u)$ domains still remains. This section introduces an algorithm that will (1) remove the remaining artifacts and (2), interpolate missing data (regularly or irregularly spaced).