| 2004 
  
{ween 
«terior 
n and 
tween 
station 
ion in 
amera 
itation 
>ssary, 
ol bar 
eters) 
nts on 
on. (2) 
linates 
d. (3) 
S just 
avoid 
ace of 
ement 
1ethod 
sary to 
ization 
ons as 
(5) 
1,2,3) 
ivation 
ice for 
'ormed 
normal 
etween 
ntation 
€) is 
hanged 
E 
nmetry 
metric 
isition. 
on and 
red and 
specific 
des can 
mn and 
SS; (2) 
rtly by 
named 
cess; (3) 
ation is 
points. 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
After field survey, all measurement data including images are 
processed by attached photogrammetry software. In this 
software, multi images based image matching (short base) is 
used instead of stereo based one in the conventional 
photogrammetry. Because geometry distortion in close-range 
photography is relatively large, traditional single-stereo 
matching which uses only two images is very difficult to meet 
the demand of matching in reliability and accuracy. Multi 
images based image matching use uses multi images and 
combines with short baseline and multi-photo perfectly solves 
the image matching and intersection accuracy problem at same 
time. This method has following characters: on one hand 
because the baseline between the neighbouring photos is 
relatively short, the geometry distortion of images is relatively 
little, thus help automatic matching; on the other hand, because 
baseline is short and multi photos are used, overlap between the 
neighbouring photos is normally very large, we can calculate 
3D coordinate with multi image intersection to achieve high 
intersection accuracy. Another feature of our terrestrial 
photogrammetry method is that control bars (shown in Fig. 4) 
located near the camera are used, which are mainly used to 
calibrate the Offset and can replace regular control points 
around or on the object to be measured. Therefore, a real 
non-contact measuring method can be implemented so that 
control points around or on the measured object can be reduced 
or completely avoided. With this technique, DTM and contour 
generation, volume measurement and 3D modeling all can be 
completed. The measurement accuracy with control bar will be 
described in next section in detail. 
  
Fig. 4 Control bar 
4 ACCURACY TEST AND ANALYSIS 
In this section, several sets of data have been used to test the 
measurement accuracy of PTSS in different situations. 
4.1 Measurement accuracy with OFFSET self-calibration 
This part aims at testing measurement accuracy when using 
OFFSET self-calibration. OFFSET self-calibration is completed 
during the process of measurement and two control bars with 
different size are used for calibration. As described in section 
3.4, bundle adjustment has been used during the process of 
calibration. In this test, a hill was measured. Six traversing 
points is arranged and the interval between neighbor traversing 
points is about 8m. 13 check points is set on the hill. Two kinds 
of test condition are used as following: 
(1) All control points and check points spread around the 
object to be measured, which is a hill here. All their coordinates 
are measured with total station. During adjustment, the 
coordinates of check points are deemed unknown. In the end, 
adjustment results are compared with observation of total station 
to check the accuracy and results are shown in Table 1. 
(2) Compared with condition 1, the only difference is that 
control points is on control bar not far from traversing point 
instead of around the object to be measured. Test results are 
shown in Table 2. Two control bars with different size are putted 
in front of each traversing point. 
From the test results, conclusion can be drawn that although the 
accuracy when control bar is used is not as high as that when 
control point around the measured object is used, it is still up to 
1/1500, high enough to meet ordinary measurement 
requirements. More important thing is that there is no need to 
put control points around of on the surface of the object to be 
measured and so photogrammetrists’ dream of “true non-contact 
measurement” becomes reality. 
Table 1 Accuracy when control points around the 
object to be measured 
  
  
  
  
  
  
  
Focus | Planar relative Z coordinate Photo 
(mm) accuracy relative accuracy | Number 
10 1/3413 1/7687 24 
30 1/4406 1/14603 105 
10&30 1/4515 1/18514 129 
  
  
Table 2 Accuracy when control points on the control 
bar not far from traversing point 
  
  
  
  
  
  
Focus | Planar relative Z coordinate Photo 
(mm) accuracy relative Number 
accuracy 
10 1/1638 1/2093 24 
30 1/2296 1/5419 105 
10&30 1/2131 1/4647 129 
  
  
  
In above test, two control bars with different size are used 
in front of each traversing point. When the number of control 
bars is reduced to one, the accuracy is shown in Table 3. 
Table 3 Accuracy when number of 
control bars is reduce to one 
  
  
  
  
  
: Planar relative Z coordinate 
focus : 
accuracy relative accuracy 
10mm 1/1365 1/1331 
30mm 1/1435 1/7876 
  
  
  
  
  
At this time the accuracy falls down, but still higher than 1/1300. 
Reducing the number of control bar can greatly save labor force 
and work time, so it can be used when the accuracy requirement 
is not very high. 
5 APPLICATION 
The usability of PTSS has been tested with two different cases. 
5.1 Case 1 
In first case, PTSS has been used to measure a stone pile in a 
concrete manufacturer where such measurement task is 
necessary and often used. The result is shown in Fig. 5.