REFERENCES: 
[1] Ackermann, F. Hochgenaue digitale Bildkorrelation, 39th 
Photogrammetric Week, Stuttgart, 1983 
[2] Makarovic, B Digitising images for automatic processing 
Tempfli, K. in photogrammetry, : ITC Journal, .' 1979-1, 
Enschede. 
[3] Turgood, J.D. Photogrammetric analysis of digital images, . 
Mikhail, E.M. ISPRS Commission III Symposium, Stockholm, 
1982 
[4] Makarovi&, B. Automatic production of DTM data by digital 
off-line technique, ISPRS Congress, Commis- 
sion II, Rio de Janeiro, 1984 
[5] Makarovi&, B. A test on compression of DIM data, ITC 
Journal, 1983-2, Enschede 
[6] Tempfli, K. Transfer Functions of interpolation 
Makarovié, B. methods, Geo-processing, 1, Amsterdam, 
1979 
ANNEX 1 Resampling 
The original digital image is assumed to be represented by a raster 
of pixels with interval d (where d equals pixel size). The corres- 
ponding Transfer Function [21 is CTESdd = f(d,ry), where a 
is a rectangular function representig a pixel (fig. 2). 
Original data can be resampled, i.e., to 'form another raster of 
pixels having a different orientation and density. The Transfer 
Function of the new raster is TE ic - f (d, r. ), where d 1s the new 
interval and r, is the new (rectangular) function. The intensity 
values of new pixels, however, are determined by interpolation 
(usually linear) from the nearest neighbouring old pixels. Hence, 
the Transfer Function for interpolated vaues in the original (old) 
raster is 
TF oid/int = f(d, interpolation) 
The combined effect of interpolation and of the new raster (interval 
d) can be represented by 
TFr esamp 7 TFoid/int'TTnew 
  
  
  
  
i 
ixel 
pase el 
| size | 
Tr d T 
fum TEE 
Figure 2: Rectangular function representing a pixel. 
350