A VIDEO-BASED INDUSTRIAL MEASUREMENT SYSTEM 
by 
Peter C. Gustafson Harry B. Handley 
Geodetic Services, Inc. Geodetic Services, Inc. 
1585 Industrial Road 1511 South Riverview Drive 
San Carlos, CA 94070 USA Melbourne, FL 32901 USA 
ABSTRACT 
Industrial measurement applications include many tasks that are time critical. While video imaging technology 
provides many advantages over film-based systems with respect to speed, the levels of accuracy and repeatability 
attainable using a film camera far exceeds that of video. However, video imaging has progressed to a point where 
acceptable accuracy levels for many industrial applications are possible. This paper discusses a video-based system 
tested on objects of 1-2 meters or more in size, demonstrating accuracies of up to 1 part in 70,000. 
1. INTRODUCTION 
Modern industry has become more and more sophisti- 
cated over the years and metrology techniques and 
requirements have contributed greatly in this growth. 
As quality assurance becomes a key to market success, 
extensive efforts are placed on reliable techniques to 
confirm dimensional integrity. Metrology systems from 
the past decade have allowed highly accurate and reli- 
able results at a level of productivity that has been 
economically attractive to the industrial end user. In 
fact, virtually all the large aerospace corporations in the 
US and many elsewhere have multiple photogramme- 
try, theodolite, CMM and other metrology systems. As 
industry becomes familiar with the current technolo- 
gies, they are ableto appreciate the advantages, andalso 
realize the shortcomings, of many measurement tech- 
niques. 
Film-based photogrammetric systems have shown sev- 
eral advantages for many different types of 
measurements. Some of these advantages include high 
accuracies, a large number of points measurable in a 
practical manner, objects in a wide variety of sizes 
possible with the same system and many others (Fraser 
& Brown, 1986; Fraser, 1992). Another of the advan- 
tages is the speed of the overall measurement and most 
particularly, of the data acquisition (quick photography 
resulting in reduced down time, etc.). However, the 
overall speed can also be a shortcoming for certain types 
of applications. Building or adjusting objects is often a 
task more suited to theodolite-based systems. Higher 
productivity and throughput would make photogram- 
metry even more widely applicable. 
Video-based photogrammetric systems are focused on 
the need for higher overall speed of measurement, most 
obviously gained through the deletion of the film devel- 
oping phase of the project. Automation of image 
measurement has improved in film-based systems as far 
as speed and reliability (Brown, 1987) but the potentials 
in video are for far greater speeds. Thus far, video- 
based systems have not gained general acceptance in 
industry. One of the main reasons for this has been the 
need for greater accuracies than those available using 
video techniques. While productivity is an advantage, 
a major part of the modern manufacturing and quality 
assurance processes require strict levels of accuracy. As 
video technology has improved, the accuracies possible 
with video-based systems have improved to a level that 
is acceptable for many applications. 
The studies discussed in this paper concentrate on the 
potentials of some higher end video equipment in terms 
of accuracy and repeatability. Three separate test sce- 
narios are discussed with points of interest from each 
highlighted. The first test involves a calibrated test field 
with a modest number of images measured several 
times. The second test simulates a full scale antenna 
survey using a particular convenient geometry. The 
final test returns to the test field approach but uses 
significantly more images per network. 
2. TEST HARDWARE 
The primary system tested includes a video camera, 
image processing boards and computer. The video 
camera used in this study is the VIDEK MEGAPLUS 
camera . The VIDEK features a CCD sensor of Class 1 
(virtually defect free) with a 1340 H x 1037 V array of 
pixels. The first and last rows of pixels as well as the first 
20 columns of pixels are shielded from light and serve 
as dark level reference to reduce the effects of dark 
current. The remaining light sensitive area of the array