hree 
| of 
  
  
  
Koistinen, Katri 
2.) Utilization of Cocentric Images 
Creation of panoramic images from cocentric image sequences has been a research topic already during several years in 
our Institute especially by Mr. Petteri Póntinen. During FJHP's excavation seasons both video camera and digital 
camera has been utilized. Image sequences has been taken by rotating the camera on the tripod 360 degrees. If the 
projection center stays approximately in one place during the rotation, the frames (or images) can be considered to be 
cocentric central projections of the surrounding world. These frames can be projected onto the cylinder surface and 
adjacent frames can be combined to a one panoramic image by using the overlaps of the frames. More details from this 
method can be found in Póntinen (2000), Haggrén et al (1999), and Mononen et al (1999). 
During excavation season 1999 cocentric images were taken especially using digital camera mounted with special 
device on the tripod. An image mosaic created from such kind of images is shown in figure 5. In this case, images are 
rectified to the image plane of the rightmost image. 
One of the advantages of image mosaics created from cocentric images is that the original resolution of every single 
image can be utilized. The need for control points is also reduced. 
2.3 Refinement of Existing DTM using Monoscopic Photogrammetry 
Currently, the tachymeter has been the only reasonable tool for 
detailed on-site mapping and collection of 3-D geometric data. 
However, the time that is necessary for performing the 
tachymeter recordings will nearly entirely reduce the time 
available for the excavations themselves. Therefore, it would 
be just beneficiary to find any image based recording solutions 
in order to reduce the total time of on-site recordings. One 
promising alternative seems to be the photogrammetric 
mapping approach based on monoscopic recordings (figure 8). 
In case a DTM already exists, detailed mapping can be based 
on single images. The coordinates of any new point P' 
identified on image will be determined as the intersection P 
between the terrain surface and the collinear image ray. The — Figure 8. Collinear intersection of new object points 
use of collinear intersections presumes that the image should on DTM using monoscopic image observations. 
be first externally oriented. This will be done by resection 
using at least four surrounding points whose 3-D coordinates 
are already known. 
  
As both the DTM and the orientation of the image may be approximate, the 3-D coordinates of the new points will be 
approximate (P) as well. However, as long as the original image observations P' are also included and stored in the 
database, the correct location of P can be determined afterwards. This will be done by a bundle adjustment combining a 
wider block of images and their observations at this location. Similarly, the DTM and the orientation data become 
refined. 
3 THREE-DIMENSIONAL DATABASE 
One goal of the FJHP is to build three-dimensional model of the archaeological project, which will combine 
geometrical and descriptive data collected during the years. The imagery collected during the years should also be 
included into the database. In addition to the model itself, the available image data will be stored into the database. The 
suitable imagery can be searched from the database and it is possible to refine the modelling based on the demand of 
various users or later needs. 
The information into the 3-D database comes from several sources and the origin of data should always be recorded. 
Data is produced using various measuring and modelling methods, which are currently developed and will be further 
improved for the archaeological purpose of the FJHP. These methods combine 3-D tachymeter measurements, 
photogrammetry using digital camera and video imagery as well as sketches, drawings and notes done by 
archaeologists. Aerial photographs and GPS have been used to create the basic geographical reference for the project. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Ainsterdam 2000. 443