International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B5. Istanbul 2004 
  
conservator confidence that no damage is occurring due to 
environmental change. 
Having established a correlation between humidity levels and 
degrees of deformation, the conservators were able to specify 
the level of environmental control that is necessary to ensure the 
long-term stability of the Retable. When the conservation 
treatment is complete, the Retable will be housed in a glass 
display case in Westminster Abbey which should provide an 
environment that would preclude the need for further 
intervention for between fifty and one hundred years. 
3. SURFACE MEASUREMENT 
Whilst it would have been possible to use laser scanning 
technology to provide a 3D surface suited to the requirements of 
the spatial database, close range photogrammetry was selected 
to provide a 3D surface model as it avoided the necessity of 
shining a laser onto the panel surface. In order to achieve dense 
measurements of discontinuous and sharply varying surfaces a 
multi-station convergent photogrammetric reconstruction 
method has been developed (Papadaki 2001). The method 
integrates target measurements and image processing algorithms 
within a convergent multi-station digital photogrammetric 
framework. The method can produce a dense cloud of accurate 
object surface points provided there is sufficient surface and 
image texture. Furthermore it has the advantage over laser 
scanning in that it can directly incorporate the automatically 
extracted 3D edge information necessary to achieve an accurate 
model of the complex surface of the Retable. 
The methodology involves the iterative densification of a sparse 
3D triangle network using the 3D locations of natural features, 
such as points and edges, which are detected and measured in 
multiple images. The iterative triangle model provides both a 
surface model and a geometric constraint for the reconstruction 
process making the system capable of processing complex 
shapes and able to account for occlusions. The resultant model 
can be used within the spatial database to drape imagery onto 
the 3D model in order to produce a realistic and measurable 
representation for the art conservator. 
3.1 Surface generation method 
Our approach to model densification deals with the problem of 
image point correspondence in a multiple image network. The 
network set-up is implemented in a digital Close Range 
Photogrammetry System (Vision Metrology System VMS 
(software), Robson, Shortis), where the retro-reflective targets, 
which form the basis of the initial sparse triangulation, have 
been measured and the imaging geometry already recovered for 
the purposes of deformation monitoring. Standard 
photogrammetric bundle adjustment procedures provide the 
information necessary to ensure calibrated image geometry and - 
agreement to a specified datum definition. 
The process of densification is initialised by applying an interest 
operator, (Forstner 1987), to create a seed point cloud. A 
multiple image point correspondence can be reliably established 
for a number of these seed points by defining criteria selected 
from a combination of radiometric and geometric properties. 
  
  
Figure 7: Interest points computed during second iteration 
Precise correspondence is subsequently achieved with a least 
squares based image patch matching routine (Gruen 1988) that 
has been modified to use information derived from comparing 
corresponding image and object triangle shapes. The number of 
conjugate points is increased in a series of stages, in which the 
matching routine is applied to all images in the network. The 
conjugate point search is optimised through the use of epipolar 
geometry constrained by common 3D bounding triangles and 
their projection into each image. This process generates a 
conjugate point list, which is filtered through photogrammetric 
intersection to produce points with three-dimensional 
coordinates at a specified level of precision. The results are 
imported into a bundle adjustment process to provide a rigorous 
evaluation of the network. 
  
  
  
  
    
   
   
   
  
  
  
  
  
  
  
   
    
  
  
  
  
    
     
    
    
   
      
   
   
    
    
  
  
  
   
  
  
  
  
  
   
    
  
  
  
  
  
  
     
  
  
  
  
   
  
  
  
  
  
  
  
   
  
     
  
    
Intern 
SEES 
  
The f 
triang 
poten 
gradi 
edges 
extra 
linke: 
simil: 
detec 
proce 
descr 
point 
then 
meth 
3.1.1 
32 
The 
dens 
This 
adju 
surf. 
deri 
mod 
ima; 
The 
inte; 
bes