(1) |

(2) |

(3) |

In the case of missing data, a diagonal weight () can be introduced that is when a missing data point is in the equation, and 1 where all data are present. This weight can also be used to eliminate edge effects caused by helical convolution.

When data are interlaced, a PEF can be estimated by spacing filter coefficients during convolution, so that they fall on known data. An example of this filter spacing is shown in Figure . The problem with this methods is that the data must be regularly sampled in all dimensions.

interlace
Examples of PEFs on interlaced data. White bins are empty, gray have data. Left: A PEF
cannot be estimated due to too much missing data. Right: The spaced PEF can be estimated on interlaced data.
Figure 1 |

10/14/2003