Darvishzadeh Varchehi, Roshanak 
  
2.1 Visual Change Detection 
In Figure 1 the methodology, which has been used in the experiment, is illustrated. In this approach, after scanning of 
the new photo, georeferencing, enhancement and other necessary corrections were done. Then the geo-corrected 
photograph was overlaid with the old digital data and changes detected using on-screen digitizing. In this case the 
detected changes were incorporated in the existing digital data. The following sections describe the procedure and the 
final results. 
2.1.1 Scanning of Photo. An oblique colour aerial photograph of 1994, which was taken with a 35-millimetre 
camera, was scanned as one band for conversion to digital format. As the scale of the photograph was almost 1/2500, a 
resolution of 150 dpi (dot per inch) was selected, which gives one pixel for every 40 centimetres on the ground. 
2.1.2 Georeferencing. After conversion of the image to the ILWIS program, the next step is georeferencing in order 
to bring the image to the coordinate system of the 1992 digital database. This is very important, as for registration of the 
image with the digital data they have to originate in the same reference. In the georeference module of ILWIS, 14 well 
distributed control points in the image and the corresponding control points in the digital data were selected. A 
projective transformation was used with a sigma of 2.38 pixels (0.95 meter), which is acceptable regarding to the scale 
of the digital data and photograph. 
  
    
New photo 
  
Cs Roof density, 
  
   
  
  
  
Digital spatial dat 
    
  
v 
  
Revised build Gal of Roof density, 
Extraction of oof area 
build.bound 
- roducts / Inp 
Legend: ** = Alternative method , | File / Database] ; / Output ; à 
Figure 1. An overview of the visual change detection approach 
  
  
  
  
  
2.1.3  Overlaying & Change Detection. After georeferencing, the aerial photograph and selection of a part of roof 
layer (which falls in the same area of photo) from the digital data, they were overlaid together. As a result the buildings 
which were somehow a change recognised. 
In ILWIS, having the aerial photograph as a backdrop image, with overlaying the buildings roofs, the changes were on- 
screen digitized in a new layer and the digital data has been updated (see Figure 2). Table 1 shows the result of change 
detection and the number of houses, which is detected. 
  
  
  
  
  
  
In The Area Of Old Data New Data Difference 
Photo (1992) (1994) absolute relative 
Number of Houses 1039 1108 +69 6.6% 
Roof Area (m ) 76000 80126 +4126 5.496 
Housing Density 35.8 38.2 +2.4 6.7% 
(building / ha ) 
  
  
  
  
  
Table 1. Comparison of old and new data in terms of roof changes 
  
314 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000.