AD i i 
(8) 
1 ones 
tion of 
1e A/D 
ic map 
raphic 
n from 
(9) 
bject's 
Objects 
r space 
P(Nida 
(10) 
(11) 
liscrete 
eighted 
ion. 
terized 
| to the 
of this 
hnikov, 
  
  
  
Yuri Knizhnikov 
3 PSYCHOPHYSIOLOGICAL ASPECTS OF OBJECT RECOGNITION UPON DISCRETE SPACE IMAGES 
A starting point at research of interpretation properties of different types of space images is the understanding of 
psychophysiological aspects of object recognition (Knizhnikov, 1997), so it is necessary to evaluate peculiar features of 
visual perception of the discrete (digital) space images. Correctness of interpretation of discrete images depends on 
quantity of pixels forming this image. This research is studied quantity of pixel indispensable for the sure interpretation 
of constructions in a small city, a settlement of an agricultural type and a settlement of a cottage type (on three discrete 
space images accordingly). 
Digitized space photoimages, obtained by KVR-1000 on scale 1 : 220 000, were used as basic materials for 
experimental interpretation. The digitization was made on a complex of INTERGRAF. The original images were 
scanned by the aperture 10 microns of the square form, and then were visualized. According to the formula (2) 
resolving power of these images was calculated as 35 mm, and sizes of pixels on terrain - 3 m. | 
On this research stage a possibility of interpretation was estimated on an example of images, which are formed by finite 
number of pixels. Such images are interpreted as integratedly percepted objects. This feature is formulated in 
gestaltpsychology (from a german word Gestalt - integrated object, integrated form). Integratedly percepted images 
have no an inside structure, but their forms are visually percepted as a unit. 
Ten separate constructions were random selected on each of three images. These constructions have minimum area, but 
sure recognized. The number of pixels, which were compounding the area of interpreted constructions, is listed in table 
2. The mean area of objects has compounded 20 pixels (piece of an image - 4 x 5 pixels). The mean area was calculated 
to better than 6,5 96 accuracy. 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Number Area of interpreted 
of objects, pixels 
object Smallcity | Settlement of | Settlement of a 
an agricultural| cottage type 
type 
1 19 20 20 
2 21 2 20 
3 19 17 16 
4 17 20 19 
5 17 23 21 
6 20 21 21 
7 22 18 16 
8 23 20 20 
9 17 19 17 
10 21 17 20 
Mean area, 20 20 20 
pixels 
Calculated + 6,5 + 6,0 + 6,0 
accuracy of (X D GE 1) en 
mean area, % 
(pixels) 
  
  
  
  
  
  
Table 2. Mean area of interpreted objects 
According to the formula (2) diagonals of the area of interpreted objects has compounded approximately five-multiple 
value of the resolution of discrete space images. The following this relationship is necessary for sure interpretation of 
integratedly percepted objects on discrete space images of defined level of generalization. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000. 505