274 
to be slowed down to the speed of human responses. If the outputs are 
generated at high speeds without human intervention the indicated conditions 
will result in extensive and costly post-editing. In both cases the 
economical gains are diminished. It is evident that even in the less 
demanding case of automation of elevation collection some additional image 
matching techniques based on criteria other than correlation of image 
density functions such as edge detection and symbolic matching that exploit 
qualitative aspects of images and consequently are less dependent ‘on gray 
level differences, have to be investigated and implemented in order to 
achieve a reliable full automation (Horn, 1983, Baker and Binford, 1982). 
For full automation, besides the on-line techniques, the off-line techniques 
could be considered. For partial automation of processes where the human 
operator monitors and aids the image matching devices or where the image 
matching device supports the actions of the human operator by performing some 
selected tasks on request (e.g. parallax measurements in orientation 
procedures) only the on-line techniques are implied. 
In more demanding cases of automation of processes where the meaning of 
patterns has to be understood and defined in a particular context, the 
development of techniques require more research closely related to the 
development in the disciplines concerned with image understanding and 
artificial intelligence. Evidently the full automation of photogrammetric 
processes represents an area whose further development would require the 
commitment of considerable resources. 
SYSTEMS FOR LAND INFORMATION DATA BASES 
In the areas traditionally served by photogrammetry the implementation of 
the data base concept in cartographic, cadastral and other land-related 
systems has significantly influenced the concepts about the objectives, the 
procedures and the configuration of photogrammetric systems for collection of 
information. , 
The primary function of these photogrammetric systems is the conversion of 
the source information registered on images into a desired, well-structured 
form which may be numerical, pictorial or graphical. Formally all the 
procedures for the achievement of a particular objective may be looked upon 
as a combination of elementary processes represented by a certain conversion 
of information. All possible elementary processes or information conversions 
are indicated in the following matrix: 
P+P P+N P»G 
N>P N>N N»G 
G>P G>N G-+G 
where P stands for the pictorial form. of. information storage or 
presentation, N for the numerical ‘or digital form and G for the graphical 
form. : 
Numerical output is the principal form for geocoded information systems 
represented by digital files stored on direct access memory devices. The 
graphical and pictorial output forms may be regarded as supporting forms for 
information storage and presentation, which may be needed either as final 
supplementary documents or as intermediate output forms required in the 
process of information conversion (Jaksic, 1981).