Page 155 - Introduction to Information Optics

1 40 2. Signal Processing with Optics

domain of a white light processor, the complex light distribution at the Fourier
domain can be shown as

"T" n I J I I i 'T' / I I J J O , IT* I
i
'
'T
— a) r I + T b I a ± — — oj ft, jg 1 + TJ a ±
TJ a, £ ± — a) r I + T b I a ± ,

T*±

where T r, T b, and T g are the Fourier transforms of T,, T b, and T g, respectively.
By proper color filtering of the smeared Fourier spectra, a true color image can
be retrieved at the output image plane, as given by

2 2
7(x, y) - T r (x, y) + T b (x, v) + T/(x, y), (2.135)

which is a superposition of three primary encoded color images.
Many of the images obtained in various scientific applications are gray-level
density images; for example, scanning electron micrographs, multispectrai-
band aerial photographic images, X-ray transparencies, infrared scanning
images, and others. However, humans can perceive details in color better than
in gray levels; in other words, a color-coded image can provide better visual
discrimination.
We now describe a density pseudocolor encoding technique for mono-
chrome images. We start by assuming that a gray-level transparency (called Tj)
is available for pseudocoloring. Using the contact printing process, a negative,
and a product (called T 2 and T 3, respectively) image transparency can be made.
It is now clear that spatial encoding onto monochrome film can be accom-
plished by the same method used for color image preservation, for which the
encoded transparency is given by

T(x, y) = /C{T,(x, y)[l + sgn(coso>jj)] + T 2(x, y)[l + sgn(cos o> 2x)]

y
+ T 3(x, y)[l + sgn(cose%x)]}~ .
Similarly, if the encoded transparency is inserted at the input white light