Yuri Knizhnikov 
  
4 PECULIAR FEATURES OF AEROSPACE GENERALISATION OF DISCRETE IMAGES 
4.1 Generalization levels of space images 
Generalization is the cartographic term, but at present, aerospace generalization is indispensable for optimization of 
representation and interpretation properties of discrete space images. The modern space images contain the 
geoinformation of many levels (Kravtsova, 1988). These levels are defined as geoinformation levels or levels of 
geoimages (Berliant, 1996). 
Quantity of geoinformation levels depends on complexity of an outline of an investigated category of objects, and also 
capabilities of images to transmit the information on objects of these levels. For example, the settlements on images can 
be shown at three levels: level of small-sized constructions inside habitation quarters (figure 3a); level of large 
buildings describing a plan of habitation quarters (figure 3b); level of habitation quarters in settlements (figure 3c). 
Object of researches can be the objects as one, and several geoinformation levels, and for each level usage of images of 
an optimum level of a generalization is expedient. Features of visual perception of objects on discrete images condition 
last position. 
   
a 
Figure 3. Discrete space images of optimal generalization levels: a) level of small-sized constructions inside habitation 
quarters; b) level of large buildings describing a plan of habitation quarters; c) level of habitation quarters in settlements 
The optimum visual perception of objects is reached on produced discrete images with calculated representational 
properties (formulas 2 and 6). It can be shown on an example of discrete images, which produced for the purposes of a 
topographic interpretation. 
4.2 Aerospace generalization for topographic interpretation 
Now, usage of space photographic images for compiling or updating of the large-scale topographic maps (1:10 000, 
1:25 000, 1:50 000) considerably extends. This process can be provided by usage of photographic images obtained by 
the KVR-1000. The representational properties of these images allow to receive copies at 20-multiple increase of the 
originals up to a scale 1:10 000. The images of such scale contain the objects even of small-sized constructions, imaged 
on a map of the same scale. 
As it was marked above, it is the first geoinformation level of images. However, according to the norms for updating a 
map to scale 1:25 000 the available information will be exuberant. According to the norms, the separate large 
constitutions describing a plan structure in habitation quarters are shown on the topographic maps of the given scale. 
That is second and more generalized geoinformation level of objects. 
Habitation quarters of cities are shown on the topographic maps of a scale 1:50 000. In a considered example it is most 
general — the third level of the geoinformation about settlements, for which one the information of initial images is 
considerably excessive. So discrete images with optimum level of a generalization are produced for interpretation of 
objects of a settlement. 
The parameters of discreting (sizes of scanning aperture and sampling rate) were calculated to supply the relationship 
between the size of a pixel and geometrical average size of interpreted objects (section 3). At this approach the 
conventional scheme of surveying in different scales is modeled, at which one the representational properties of 
surveying systems provide a different values of the resolution on terrain and photographic contrasts of original images 
(figure 4). 
  
506 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.