i
end
image matching and data extraction. An accompany-
ing paper* describes the basic principles of optical
correlators and discusses their application to photo-
grammetric data processing, emphasizing the capabili-
ties of these devices to process large image areas
simultaneously. Since optical correlators avoid the
process of scanning the imagery and converting it to a
time-domain signal, they overcome much of the
inherent speed-sensitivity limitation of flying-spot
scanners. Figure 124 shows a comparison of the
theoretical signal-to-noise ratio of comparable elec-
tronic and optical correlators for low-contrast im-
agery of high spatial bandwidth. Significantly, the
optical correlator performance is independent of
spatial frequency (within the frequency limit of its
optical aperture, 280 line-pairs per millimeter), and
the signal-to-noise ratio can be greatly superior when
a laser power of greater than a few hundred milliwatts
is used. A limitation of the present optical correlators
is the difficulty of compensating for image distortion.
The present systems compensate only for first-order
distortion and require additional optical elements to
introduce rotation and anamorphism. While there are
a number of potential techniques for improving and
simplifying distortion correction, much development
remains to be done in this area.
300
Space- Bandwidth Limit
100 1— : Region Ci
Optical With Higher Power
Correlator
(100 MW Laser)
36 +
z
o
2
S
t€ 101—
$8
© Electronic
T Correlator
>
$ s—
o
-
3
=
3
©
1 dl
Contrast
031— Modulation = 4%
Average Density
0 1 2 3
o11— | | —
0 0.1 0.01 0.001
Average Intensity Transmittarice T
Figure 12 Signal-to-Noise Ratio Performance of the
Optical and Electronic Correlators
AUTOMATION IN COMPILATION
Scanning
Signals Deflection
Image Fourier Image x. wv and
Plane
Plane Transform
Image
Intensifi
Input
Photograph
Figure 13 Camera-Tube Scanning System
Another area of interest is the application of
camera tubes to image scanning. The image dissector
is of particular interest because it avoids the require-
ment of the vidicon and the image orthicon for a
highly regular scanning pattern and is fully as flexible
as the flying-spot scanner in this respect. Figure 13
illustrates some of the advantages of a camera tube
system. Spatial filtering, employed before the scan-
ning process, can be used to improve the contrast of
the scanned image and to emphasize components of
the imagery which contribute to the x-parallax deter-
mination. The image intensifier is needed, in general,
to provide sufficient light input to the image dissector
to achieve a high signal-to-noise ratio. In the optimum
system, the image intensifier and image dissector
would also provide integration or smoothing of the
electronic image analogous to that provided in the
vidicon and image orthicon. The major barrier to the
application of these techniques at the present time is
that the basic components, the image intensifier and
the image dissector, are not yet fully developed for
this application. However, recent work at Bendix
Research Laboratories in the application of the Chan-
neltron®* miniature electron multipliers to these
problems shows substantial promise.
Other potential flying-spot scanner replacements
might be mentioned, for example, the deflected laser
beam, but each has fundamental limitations which are
yet to be overcome. At present there is no alternative
available which can compete with the flying-spot
scanner and electronic correlator on all fronts.
*S. J. Krulikoski et al, “Coherent Optical Parallel Processing.”
1 This figure has been taken from an acompanying paper, “A Com-
parison of Optical and Electronic Correlation Techniques," by D. C.
Kowalski.
(Q* Registered trademark, The Bendix Corporation.
11