Accuracy]
(mm) j
0.022 3:
: Actual accuracy
0018. dosi cw eso Cause .
0.014 | Theoretical accuracy
0.010 | Estimated accuracy
0.006 1—L—] | |
8.10.30 Numbé Stations 120
Figure 5
Graph of Accuracy vs. # of Stations
as before and the results reported are the closures of
triangulation (Table 7) and the estimated and actual
accuracies (Table 8). Actual accuracies are determined
from the RMS residuals of a three dimensional coordi-
nate transformation into the film-based measurements.
The accuracy results achieved are actually quite good.
The range of relative precision is from 1:36,000 to
1:51,000 of the major dimension of the object with these
Closures for Each of Eight - 8 Station Runs
Test # s i s : s s Rabo iid
1 .28 25 27 1/25
2 22 20 21 1/32
3 28 23 26 1/26
4 25 20 23 1/30
5 27 -19 23 1/30
6 24 22 23 1/30
7 23 21 22 1/31
8 22 21 22 1/31
Table 7
Estimated and Actual accuracies for
Each of Eight - 8 Station Tests
Estimated Actual
Test # VXYZ VXYZ
1 0.025 0.023
2 0.025 0.033
3 0.022 0.023
4 0.021 0.031
5 0.025 0.023
6 0.025 0.033
7 0.022 0.023
8 0.021 0.031
Table 8
8 station versions. However, as seen from the previous
section there is a considerable advantage to the mea-
surement of multiple images. This advantage is expected
to improve significantly when suspected unknown
systematic errors are traced and properly modeled.
4. CONCLUDING REMARKS
Specific problems still lie in the stability of the lens for
the video camera tested (Gustafson. 1991). After isolat-
ing the lens from the strobe, lens element movement is
still noticeable. At the time of the submittal , another
lens had not yet been tested to confirm whether the
problems were from a single lens or from the lens type
which may be the source of the systematic errors
discussed in 3.3. Additionally, many of the results
suggest that there may be cases wherein models typi-
cally used to correct for systematic effects in most film
cameras do not completely model systematic effects in
the video camera used. As pointed out previously, there
is also an expectation of improvement using improved
thresholding techniques.
The general trend of the results discussed is good.
Accuracies at the level of use in industrial environment
have been obtained and confirmed (up to 1 part in
70,000). There certainly appears to be room to improve
but with the results obtained to date, studies will con-
tinue.
Acknowledgments
The equipment used in this series of tests was provided by Northrop
Corporation of Pico Rivera, California. Their continuing support has been
appreciated, particularly in the search for applications. The authors would
also like to thank their coworkers at GSI, without whose help the tests
discussed, and this paper, would not have been possible.
References
BROWN, D.C. (1987): “AutoSet, An Automated Monocomparator Opti-
mized for Industrial Photogrammetry.” Presented Paper, International
Conference and Workshop on Analytical Instrumentation, Phoenix, Arizona,
November.
FRASER, C.S. (1992): “A Summary of the Industrial Applications of
Photogrammetry.” Invited Paper to the First Australian Photogrammetric
Conference, Sydney, November.
FRASER, C.S. and D.C. BROWN (1986): “Industrial Photogrammetry:
New Developments and Recent Applications.” The Photogrammetric Record,
12(68): 197-217.
GUSTAFSON, P. C. (1991): “Accuracy/Repeatability Test for a Video
Photogrammetric Measurement.” Industrial Vision Metrology, SPIE Pro-
ceedings, Vol. 1526, pp. 36-41.
TRINDER, J.C. (1989): “Precision of Digital Target Location.” Photogram-
metric Engineering and Remote Sensing, Vol. 55, No. 6, pp. 883-886.
SHORTIS, M.R., A.W. BURNER, W.L. SNOW AND W.K. GOAD
(1991): "Calibration Tests of Industrial and Scientific CCD Cameras."
Presented Paper, Coordinate Measurement Systems Committee Meeting,
Seattle, Washington, October.
