Notes on Photogrammetry for measurement of antennas


Bryan Butler, April 1999.

One of the techniques which could possibly be used to measure the ALMA antennas is that of "photogrammetry" - specifically "digital photogrammetry." In the technique of digital photogrammetry, a specially built photogrammetric camera is used to photograph targets placed on any surface from several locations, at several angles. Three dimensional coordinate data are determined from the two dimensional photographs via optical triangulation. This is called "convergent photogrammetry", to distinguish from "stereo photogrammetry", where only two camera locations are used, with the camera axes parallel. This technique has been increasingly used in industrial applications, including antenna reflector measurement (Fraser 1992, 1993). The technology is referred to by several names including non-topographic photogrammetry, terrestrial photogrammetry, close-range photogrammetry, and industrial photogrammetry. Originally, the photographs were recorded on thick glass plates, since glass was more stable than film. These stability problems were eventually overcome, and until recently the photographs were recorded on film. Recently the trend has been to record the data digitally on a CCD, hence the name "digital" photogrammetry (also commonly called video-grammetry). The concepts remain the same, however.

The accuracy obtained is strictly proportional to the size of the imaged surface. Current typical digital system accuracy is about 100000:1. The accuracy obtained may be improved by "mosaicing" (sometimes called "subsectioning") the imaged surface. Accuracies of 250000:1 are typically obtained, and even better results have been obtained in special circumstances. Note that an accuracy of 250000:1 gives about 50 microns rms for the 12 m ALMA reflector surfaces. Note also that this accuracy gives about 3 microns rms for measurement of a 0.7 m subreflector.

Digital photogrammetry systems are currently available off-the-shelf, with little to no modification. The current industry leader in providing these systems is Geodetic Services, Inc. (GSI), in Melbourne Florida. GSI will provide a digital system which includes the photogrammetric camera, strobe, notebook computer, and proprietary software for about $150000. The entire digital system fits into a carrying case, with total weight of about 15 pounds (it will fit under an airplane seat).

NRAO has now had two experiences with such measurement systems. The first was a measurement of the subreflector for the Green Bank Telescope (GBT). This subreflector is 8 meters in diameter, and was measured with a film system. The measurement accuracies for the many different measurements were between 0.4 and 1 thousandth of an inch (10 to 25 microns), i.e., about 500000:1. This is about what is expected, as the film systems are typically about a factor of 2 more accurate than the digital systems. The second experience was a proof of concept measurement of a VLA antenna with a digital photogrammetry system. NRAO contracted to have GSI come to the VLA site and fully measure a VLA antenna primary reflector surface, the surface of the subreflector on that antenna, and the surface of the spare VLBA subreflector which is currently stored at the VLA site. Examination of the data from this visit is not complete, but initial results are very encouraging. Repeatability accuracy for the 25 m primary reflector surface was around 70 microns in the best cases. This is about 350000:1 - a bit better than typical. Repeatability accuracy for the VLA subreflector was about 15 microns, and for the VLBA subreflector was between 20 and 25 microns, or roughly 170000:1. The measurement accuracy is worse for the subreflectors because the convex shape is not ideal for photographing (less duplicate coverage of targets).

In terms of accuracy, it seems that a digital photogrammetry system will not meet the requirements for the final measurement of the primary reflector surfaces of the ALMA antennas. Even given a 350000:1 measurement, the achieved accuracy is only 35 microns. The surface of the primary reflector is supposed to have an rms of 25 microns, which means that it will have to be measured to about 15 microns rms - more than a factor of 2 better than the digitial photogrammetry system might achieve. However, it might be possible to use the digital photogrammetry system for the initial set of the antenna surface. Target placement would be a complication, but is probably not an insurmountable problem. The digital photogrammetry system might be sufficient to measure the subreflector surface, however. A 170000:1 measurement would give 4 microns rms on a 0.7 m subreflector, which should be sufficient. Note also that the digital photogrammetry system might be used to measure other parts of the antenna structure.


Fraser, C.S., J. Photogramm. Rem. Sens., 48, 12-23, 1993
Fraser, C.S., Photogramm. Eng. Rem. Sens., 58, 305-310, 1992