scenes contain large continuous areas with edges, they contain a large
D.C. and low frequency component. Most images have spectra where the
magnitude tends to drop off drastically with increasing frequency.
The energy in the low frequencies may be several orders of magnitude
larger than the high frequencies. However, it is the high frequencies
which contain the useful information in separating the desired target
from false targets. A practical problem with holography is the
dynamic range to be recorded. Film cannot typically induce more than
two or three orders of magnitude of dynamic range. To record a
hologram of the Fourier transform, the film must accurately record the
entire dynamic range of the transformed image. If the dynamic range
of the transformed image is too large, the film cannot record the
Fourier transform linearly and the correlation is not ideal. The film
non-linearity will emphasize some frequencies and attenuate others.
The correlation signal-to-noise ratio will suffer if important
frequency components are attenuated. To reduce the dynamic range of
the transformed image and allow linear recording on the hologram, the
useless frequencies in the image should be eliminated. Because the
low frequencies contain most of the image energy but little of the
information, their omission considerably reduces the dynamic range
with little effect on the correlation except to reduce the overall
light through the hologram.
To determine which frequencies are important in target
discrimination involves considerably more work than can be considered
here. In general, a set of target images and a set of non-target
images can be compared on a large digital computer to determine which
frequencies appear most in the desired target. This requires a large