In these expressions are
<S X ’ S y ’ S z — standard errors of the final coordinates
D = distance from the camera to the object
c = the camera constant under water
s 0 = standard error of unit weight ofy-parallax measurement on the image scale
If the formulas above are used even though the residuals are not normally distributed an approximate
estimation of the final accuracy is found. This gives for the Hasselblad camera tested (c = 54 mm and
s 0 = 38 urn)
S x — S y — 0.008 D meter (= 0.08 m for D — 10 m)
S z = 0.019 D meter (= 0.19 m for D = 10 m)
and for the tested Rolleiflex camera (c = 103 mm, s 0 = 23 urn)
S x = Sy = 0.003 D meter (== 0.03 m for D = 10 m)
S z = 0.011 D meter (= 0.11 m for D — 10 m)
1.84 Tests of Ballistic Cameras
Very interesting information has been received on the accuracy of photographs from ballistic cameras
from Mr G. H. Rosenfield, Missile Test Project, Patrick Air Force Base, Florida. Mr. Rosenfields writes:
’During the past years, the Ballistic Camera System has been considered to be the accuracy
standard at AMB with a prime mission of evaluation and calibration of other instrumentation systems,
both optical and electronic. The planned acquisition of an electronic measuring system known as
MISTBAM, however, necessitated greater accuracies for the Ballistic Camera Systems if MISTBAM
was to be evaluated. The result was the Ballistic Camera Accuracy Review (BACAR) Committee set
up to review the present status of the System, to improve the accuracy of the System in order to meet
MISTRAM evaluation requirements, to determine maximum theoretical accuracy capability of the
System, and, having reached the practical maximum accuracy capability, to establish a program to
preserve this and to develop refinements leading to a further accuracy improvement.
The BACAR Group is composed of RCA employees whose major work covers a great variety of
interests and knowledge. Such was needed if a complete analysis of the problem was to be undertaken.
The group needed to study facilities, cameras, photographic plates, plate reading equipment, data
reduction techniques, mathematical models, star catalogues, etc. The operation of the plan was broken
down into two phases, the first to deal with the immediate problems of MISTRAM evaluation and
actual theoretical capabilities of the present System. Phase II is to be the improvement on a con
tinuing basis. The project was undertaken in late 1961 and to date the results have been impressive.
For example, the standard error for camera orientation of 4.4 /urn may be reduced to 2.3 if all error
source calibration are utilized and if proper care is taken in the entire operation.”
The value 2.3 ^m mentioned is the estimated standard error of image coordinates in plates from
BC-300 cameras and may be considered the best possible accuracy which can be expected. It is of
particular interest to note the estimate of error components from which the standard error of image
coordinates mentioned is calculated.
Standard error
s
um
s 2
Identification
1.5
2.25
Stereocomparator
1.0
1.0
Emulsion
1.0
1.0
Star catalogue
1.0
1.0
Refraction
0.3
0.09
(5.34)9* — 2.3 ^m image coordinate
26