For automatic operation, analytical systems can be converted to 
hybrid systems. Examples are the Bendix AS 11 BX (Drum, 1978), Kern 
DSR 11 (Cogan, 1984), Zeiss C 100 Planicomp (Zeiss, 1987), and 
Helava Assoc Inc, HAI DCCS (Helava, 1986). 
In at with analytical stereoplotters, prepositioning and point 
transfer can be digitally controlled; measurements of the new points 
are manual. The operation is relatively slow and sometimes 
error-prone. On-line gross error detection/removal can be included. 
The required storage capacity is modest. 
In hybrid and digital systems, at can be largely automated. The 
operator initiates and supervises the process, accepts/rejects 
intermediate and final results, helps or overrides the system, and 
ends the operation. 
Automatic selection of the tie points is done by approximate 
prepositioning and the choice of a small image window on the master 
image by using an "interest operator". Then the selected windows are 
matched with the conjugate segments on the overlapping images. Image 
matching is usually from coarse to fine, and it can be simultaneous 
for all conjugate segments with the selected window on the master 
image. The matching algorithm can incorporate the collinearity 
constraint and the least squares fit, and it can be extended by the 
quality control. 
In hybrid systems, small image windows, centered on the points of 
interest, need to be digitized and stored. In digital systems, 
however, full images have to be digitized, which requires great 
storage capacity. 
Automation of at leads to an increased production rate and a more 
homogeneous performance. The limitations are a high price for system 
acquisition, uncertainty about the support, and the risk of early 
obsolescence because the techniques are changing fast. 
Installation of an automated hybrid or digital system for at seems 
justified when the amount of at to be carried out is great, e.g., 
for vaste territories and/or when the system is also used in other 
production. 
5.2 Terrain relief modelling 
Both system types are suitable for the commonly applied manual and 
semi-automatic techniques for relief modelling. 
Analytical systems lend themselves particularly for semi-automatic 
digital modelling, such as by a regular point grid, progressive 
sampling and composite sampling (Makarovic, 1977). These techniques 
are suitable for fine (detailed) modelling, which requires 
interpretation of terrain at each point to be measured. 
For medium-fine and coarse relief modelling, however, automatic 
techniques can be more efficient. To this end there are two 
314