5A-5-2 
Photogrammetric traverse 
Suppose a sequence of stereo images taken by a pair of 
cameras from the top of a vehicle that moves along a street 
or road. In a MMS, the images are usually positioned by 
GPS and oriented by an INS (Inertial Navigation System). 
When GPS data are not available (and this is quite common 
in urban areas), INS data fills in the blank positions. And 
when one does not have the INS to do it, orientation is 
given by a bundle block adjustment. This is what we have 
called photogrammetric traverse (Silva & Oliveira, 1998). 
Figure 1 shows that an object point or detail on the street, 
for example, may be clearly seen in two or three stereo 
bases and then in four or six images. Simple, double or 
even multiple photogrammetric intersection can compute 
the spatial object coordinates. Theoretically, the closest 
stereo-base delivers the highest accuracy and the far base 
the lowest accuracy for an object point when the 
photogrammetric intersection is computed separately for 
each base. When the computation takes in account multiple 
intersections the final accuracy is diminished mostly by the 
far bases’ data due to unfavorable geometry and minor 
quality observations. Edmundson&Novak (1992), Habib 
(1994) and Silva (1996) have treated sequences of digital 
images. If the photocoordinates are obtained in a manual 
(visual) measurement process, of course, this is restricted to 
a non real time system. 
Figure 1 - Mapping the object points in a sequence of 
image pairs. 
3 THE WORK IN THE FIELD 
For convenience the streets circumscribing two blocks near 
the campus were chosen to be the test field (fig. 2). The 
planimetric positions to be occupied by the digital camera 
(Kodak DC40) were marked on the streets. The camera 
stations were separated by 1 m across the streets 
representing the camera stereo-base. The distances between 
consecutive stereo-bases varied from 5 to 20 m, 
approximately, along the streets according to the situation, 
comers and middle (between two comers) of the street, 
respectively. The advances of the stereo-base represented 
the vehicle in its trip along the streets. 
The GPS antenna of the rover receiver (Ashtech Reliance: 
carrier LI, code C/A) were put on a tripod 1.30 m above 
the ground (pretending the camera perspective center 
position on the top of the vehicle). Of course, in the true 
life 2.0 meters high would be more realistic, but that would 
bring more practical difficulties in this particular 
simulation project. The planned precision for its position 
was 10 cm, based on the estimated mismatch about 5 cm 
between the camera perspective center and the antenna 
center. Differential GPS (DGPS) was used having an 
Ashtech ZXII (carrier LI, L2; codes C/A, P, Y) on the 
reference station. The base receiver (ZXII) was 
programmed to collect data every 5 sec, from the beginning 
to the end of the surveying, while the rover receiver 
(Reliance) collected data every second for one minute per 
camera station, totaling 76 points in the surveying section. 
Later, in a second surveying section, 6 checkpoints were 
collected in the same way. These check points were 
selected according to their definition both in the field and 
in the images. 
A Kodak DC-40 digital camera was used to take color 
images. It is a CCD still frame camera, 756 x 504 pixels 
(pixel size approximately 45 pm), 24-bit color (RGB), and 
fixed focus. A previous calibration determined the 
following parameter: calibrated focal length = 46.891 
(±0.066) mm, x 0 = -0.133 (±0.052) mm, y 0 = +0.208 
(±0.068) mm and kj = 0.000063 (±10' 6 ) (Tommaselli & 
Nobrega, 1997). 
The camera was fixed on the tripod just after removing the 
GPS antenna. Its axis was made approximately parallel to 
the street longitudinal axis and the second picture of the 
same stereo-base (1 meter to the right) was taken in the 
same way making the (approximately) normal case of 
terrestrial photogrammetry. 
4 THE WORK IN THE LAB 
All data collected in the field was downloaded to a Pentium 
II computer and processed by DGPS Reliance. The 
rejection criterion of GPS data processing indicated that 42 
of the 76 camera stations had to be reoccupied to give the 
planned precision (10 cm) probably due to the buildings 
and trees. WGS84 (World Geodetic System 1984) 
coordinates were transformed to SAD69 (South American 
Datum 1969) and then to UTM (Universal Transverse 
Mercator, X = 51° W). The estimated precision of the 76 
station coordinates (perspective centers) surveyed by GPS 
were expressed by the following standard deviations: 
2.5cm, 3.6cm, and 6.1cm in E, N, and h coordinates, 
respectively.