In theory, a sampled imager is optimized by removing all aliasing. One applies presample and postsample filters at half the sample frequency; one avoids inband spurious response by removing any frequency content beyond the halfsample frequency before sampling; and one avoids out-of-band spurious response by removing any frequency content beyond the half-sample frequency on the display.
It is possible to create a postsample filter that is close to ideal. How close to ideal and how to do it are the main topics of this chapter. If the correct interpolation function is used, a display with multiple pixels for each sensor sample provides a good postfilter.
Unfortunately, adding preblur beyond that provided by the optics and detector is as likely to hurt performance as to help. There are multiple reasons for this. One is that an "ideal" prefilter in the same sense as interpolation is not possible. The postfilter is applied by digital manipulation of the samples. Once the samples are taken, it is too late to apply a prefilter.
Adding preblur by using slower optics, defocusing, adding fiber optic plates, or other optical schemes degrades in-band MTF. Removing frequency content beyond the half-sample frequency is good. Blurring the image to do it is probably counterproductive.
A third problem with adding preblur is that in-band aliasing might not be all that important, anyway. Blur is undesirable. Noise is undesirable. Aliasing might be undesirable, depending on the target, the task, the imager, and the range.
Chapters 6, 7, and 8 discuss performance evaluation. The trade-off between blur and aliasing can be made once a scenario is selected and target contrast, size, and range are known. There are certainly times when degrading optical blur improves range performance. Unfortunately, that kind of trade-off is scenario dependent.
Chapter 5 discusses interlace and dither. Both remove in-band aliasing by providing better sampling of the scene. In the remainder of this chapter, the benefit of display interpolation is discussed.
