2. 
  
TEST AREA 
Because of cost and time reasons, an old large scale 
(1:4000 ) photogrammetric block over the small Finnish 
town of Forssa was selected for the test. The 
photography had been made for practical mapping 
purposes in 1989, and has already been measured and 
adjusted with success. The total block consists of 60 
images in 7 strips with 60 % side and end laps. Due to 
problems with a high amount of image data, a small 
subblock of 4 strips and 28 images ( 7 in each strip ) was 
chosen for the test. The flight parameters are shown in 
Table 1, and the geometrical configuration of the block is 
presented in Figure 1. 
Time of photography 3.5.1989 11:40 
Flying height above ground 600 m 
Scale of photography 1:4000 
Camera Wild RC 20/23 
Principal distance 153.19 mm 
Film positive colour 
Number of strips 4 
Number of images/strip 7 
End lap 60 % 
Side lap 34 % ( 24% - 49% ) 
Table 1. Flight parameters of test block Forssa. 
2.1 Ground points 
The area was covered by a dense geodetic network, 
which was signalised using cross-shaped targets, the 
arms of which were 60 cm x 10 cm. Most of the points 
were measured using traverse measurements, and the 
coordinate accuracy of the signals was estimated to be 
below 15 mm in all three directions. 
The control points for dense ( 14 XYZ-points ) and for 
sparse ( 4 XYZ- and 4 Z-points ) control, as well as the 
check points ( 50 XYZ- and 41 XY-points ), were selected 
from this group. 
  
  
| 
TF 
Cyr TT 
ET 
| |) L | 
Lx y e 
| 
| 
| 
E mem EEE SEM 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
he 
| ]| 
= 
— 
  
Figure 2 Control points in dense control 
configuration. All 14 points are XYZ-points. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
  
2.2 Image material 
The block was scanned in Landesvermessungsamt 
Baden-Württenberg in Stuttgart/ Germany using a Zeiss 
PS1 scanner in black-and-white mode with 15 um pixel 
size. The scanning succeeded quite well except for some 
minor disturbances in image quality. Altogether, this very 
small block occupied almost 7 gigabytes demonstrating 
the problems of data storage. However, it was decided 
not to use any compression in this test, even if it seems to 
be a necessity in large blocks. Another image data set of 
30 um pixel size was created averaging every 2x2 pixel 
window. 
3. TASK OF THE PARTICIPANTS 
The task of the participants was to perform 
photogrammetric aerotriangulation with both data sets. 
The aim of the test was to investigate all varieties of the 
item, so there were no instructions concerning the 
measuring process. The choice of block adjustment 
software was also free. The blocks were adjusted using 
dense ( 14 XYZ) and sparse ( 4 XYZ and 4 Z ) control as 
well as with and without additional parameters. 
Some participants made the measurements using original 
film diapositives. Their results served as a comparison to 
contemporary photogrammetry. 
4. MEASUREMENTS OF THE PARTICIPANTS 
In all 13 participants, which have been listed below, have 
submitted results to the pilot center. 
Agricultural University of Norway Norway 
École Polytechnique Fédérale de Lausanne Switzerland 
Eidgenóssische Technische Hochschule Switzerland 
Finnish Geodetic Institute Finland 
Helsinki University of Technology Finland 
Institut Géographique National France 
Kungliga Tekniska Hógskolan Sweden 
National Land Survey of Sweden Sweden 
Technical University Munich Germany 
Technische Universität Wien Austria 
Universite Laval Canada 
University of Stuttgart Germany 
University of Trondheim Norway 
The participants made 155 different block adjustments. 
They were based on measurements, which can be 
divided into three main groups: 
1. Completely visual methods ( VM ). Measurements 
are made on a screen in the same way as on an 
analytical plotter. They can be made in stereoscopic or 
monoscopic mode. When comparing to conventional 
methods, the advantages are the flexibility in treatment 
and display of images. The disadvantages are mainly 
related to a decrease in image quality. 
2. Semiautomatic methods ( SM ). In this group 
automatic image matching has been used in some form in 
different tasks of image block measurement. The most 
common way is to match any kinds of homologous points 
on multiple images. Another automated task is the 
    
   
   
   
    
    
   
    
    
   
   
    
      
       
     
   
    
   
   
     
   
    
   
   
    
  
   
   
   
   
   
  
   
   
    
     
   
    
   
   
   
     
   
    
measur 
templat 
approxi 
or gain: 
3. Autc 
compai 
selectic 
image 
methoc 
at leas 
human 
In the 
groupe 
Within 
and tyr 
École 
The m 
Image: 
comme 
measu 
adjustr 
param 
The £ 
measu 
EPFL: 
EPFL : 
EPFL. 
EPFL 
EPFL 
Eidge 
( ETH 
The m 
The s 
meast 
using 
Finnis 
The 
photos 
used 
Cross- 
differe 
1. Cor 
2. Diff 
3. Difi 
groun 
4. Use 
FGI 1