943 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part B5. Beijing 2008 
the numbers of the image pixels belongs to the targets decrease 
as the distance of the target increases. On every position the 
integration time was set up from 200ps to 51.2ms in 2000|is 
intervals, except of the distances 0.10m, 0.25m, 0.50m and 
0.70m. Furthermore, air temperature was controlled with an air 
conditioning system in the laboratory in order to assure almost 
the same air temperature conditioning for each measurement as 
much as we could. 
Figure 11 presents the standard deviation of the first order plane 
fitting with various distances of the target from the range 
camera with respect to different integration time. First, a greater 
dependency of the measurement precision is observed on the 
integration time in the shorter distance, 25cm to 70cm, as 
shown in Figure 11a. In this range of the distance, i.e. smaller 
than 70cm, the maximum difference in the measurement 
precision is about 600%. In addition, in the medium range of 
the distance, lm to 1.5m, this effect is much smaller but the 
maximum difference in the measurement precision is still as 
large as 150% as shown in Figure 10b. Third, as we can 
observed in Figure lie, the importance of an optimal 
integration time is much less important in a long distance of the 
target from the range camera. In other words, there is a large 
size of basin in the integration time, about 30ms, where any 
integration time provides as large as the standard deviation of 
the residual, 1cm. 
Pointcloud 
(b) Tjntegjation 51.2ms 
Figure 10. Measured 3D point clouds with different integration 
time (T[ nte g ra tj on ) 
(a) Short distances 
(b) Medium distances 
0 20 40 
Integration time [ms] 
■1.75 m 
■2.5 m 
3.5 m 
•2 m 
2.75 m 
4 m 
(b) Distances between 1.75m and 3m 
Figure 11. Standard deviation of the residuals vs. the integration 
time with various distances of the target. 
1.3 Amplitude threshold 
The last parameter to be tested is the amplitude threshold. The 
amplitude of a range camera is measured from the amount of 
emitted light that is reflected back on the pixel of the camera, 
which means it can qualify the measurement for each pixel. A 
pixel of the camera’s detector with a lower amplitude value 
than the threshold can be filtered out by a threshold value. 
There is a lot of background illumination, which could 
influence the distance measurement of the target, especially at 
the comers of an object. As an example, in Figure 12b, whole 
background was eliminated (blue colour in Figure 12a) and only 
the target with its tripod remained.