The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
1063 
• Tier III < N/A 
In the document produced by the DPRG and BMGS, entitled 
Small & Medium Format Digital Camera Specifications, 
precise details are given in terms of the relationship of the 
ground sampling distance (GSD), flying height, camera 
specifications, and the above categories. 
4.2 Standards and Specifications for Digital Camera 
Stability 
The estimated IOP from temporal calibration sessions must 
undergo stability analysis to evaluate the degree of similarity 
between reconstructed bundles. When the stability analysis is 
performed according to section 3, the RMSE offsel value is 
computed to express the degree of similarity between the 
bundles from two sets of IOPs. The cameras must meet the 
following specifications to be deemed stable. 
• Tier I RMSE offset < 1 Pixel 
• Tier II RMSE offset <1.5 Pixels 
. Tier III RMSE offset : N/A 
5. EXPERIMENTAL RESULTS 
The cameras tested in this research were two Prosilica GC1020 
CCD cameras with Gigabit Ethernet interface (Figure 6). These 
sensitive cameras can collect 33 frames per second, are based 
on the Sony ICX204AL CCD sensor, have a resolution of 
1024x768, and a pixel size of 4.65pm (Prosilica Inc., 2007). 
The camera lenses used were two 6mm Pentax TV lenses. The 
camera and lens specifications are summarized in Table 1. The 
camera calibration test field used to perform all calibrations of 
the Prosilica cameras is roughly 5x5 m 2 and contains both line 
features and signalized targets. 
Sensor Type 
Sony ICX204AL CCD 
Image Resolution 
1024 x 768 pixels 
Pixel Size 
4.65pm x 4.65pm 
Size 
33mm (height) x 46mm 
(width) x 59mm (length) 
Lens 
6mm Pentax TV Lens 
Table 1: Camera and lens specifications 
Figure 6: Prosilica GC1020 cameras 
5.1 Calibration Results 
The Prosilica cameras were calibrated by the DPRG at the 
University of Calgary in March and April of 2008. Each camera 
was calibrated three times over the course of a month. Between 
each calibration session, the lenses were removed and re 
attached. The calibration results for the first camera (herein 
referred to as camera 1) are shown in Tables 2, 3, and 4. The 
IOP shown in the tables, however, cannot be compared directly 
in order to determine stability due to correlation among 
parameters. The camera stability analysis, which will determine 
if these cameras have remained stable over time, will be 
performed in Section 5.2. 
x„(mm) 
0.1774 
ox D (mm) 
0.0017 
y„ (mm) 
-0.0977 
oy„ (mm) 
0.0018 
c (mm) 
6.1377 
oc (mm) 
0.0027 
kiimm' 1 ) 
-5.1259e-03 
ok] (mm' 1 ) 
2.028 le-05 
Table 2: IOP from the Camera 1, calibration #1 
x„(mm) 
0.1793 
ox D (mm) 
0.0019 
y D (mm) 
-0.0862 
oy n (mm) 
0.0018 
c (mm) 
6.1400 
oc (mm) 
0.0026 
kiimm' 1 ) 
-5.075le-03 
ok] (mm' 1 ) 
2.1265e-05 
Table 3: IOP from Camera 1, calibration #2 
Xn(mm) 
0.1646 
ox n (mm) 
0.0020 
y»(mm) 
-0.0837 
oy n (mm) 
0.0020 
c (mm) 
6.1390 
oc (mm) 
0.0027 
kiimm' 1 ) 
-5.111le-03 
ok] (mm" 1 ) 
2.1276e-05 
Table 4: IOP from Camera 1, calibration #3 
5.2 Stability Results 
Stability analysis was performed on the two Prosilica cameras, 
according to the ROT method outlined in Section 3.2. This 
method was chosen because under the terrestrial applications 
being considered in this work, the only constraints likely to be 
imposed on the cameras are that the camera positions may be 
restricted. In some cases there may be no restrictions imposed 
on the cameras in terms of position and orientation, and in such 
cases the SPR method could be employed. However, since the 
ROT method gives more conservative results then the SPR 
method, this stability analysis method was chosen for use in this 
research. To demonstrate the stability of the utilized cameras, 
we evaluated the degree of similarity between the reconstructed 
bundles from the various calibrations, and the RMSE offset values 
for the two cameras are shown in Table 5. It is evident that 
these RMSE offset values are no greater than one pixel size 
(0.00465mm), which indicates excellent stability of the tested 
cameras. 
Data sets: 
RMSE For 
Camera 1 (mm) 
RMSE For 
Camera 2 (mm) 
Set 1 vs. Set 2 
0.0006 
0.0039 
Set 1 vs. Set 3 
0.0004 
0.0015 
Set 2 vs. Set 3 
0.0005 
0.0025 
Table 5: Camera stability results 
5.3 Analysis of Distortion Model Equivalence 
The stability analysis tool was also used for the comparison of 
IOP derived using different distortion models. In this research, 
the equivalence between three distortion models was 
investigated. These models include: Krauss (Equation 1), 
SMAC (Equation 2), and PCI (Equation 3) distortion models. In