278 279 
precise, well-controlled photogrammetric cameras from airborne and space 
  
platforms. Hence most of the information to be processed by a digital system 
has to be converted into digital form by scanners. Therefore the analysis of 
scanning and quantizing techniques with respect to scanning speeds and 
d minimal entropy of information is an essential task to be addressed. 
y 
© The next critical ares of concern "is the storage of the digitized 
f information. Assuming that 256 levels of gray are satisfactory each picture 
d element (pixel) requires one byte. If a standard 23cm x 23cm photograph is 
r converted into an array of 10um pixels (which already implies some loss of 
e information) the resulting array would require a storage space of 529 Mbytes. 
. This indicates not only the enormous requirements for storage space on 
y external memory devices but also the problems related to the access times and 
a the size of necessary random access memory. 
d 
It is evident that a photogrammetric digital image processing system should 
be able in the first place, to perform all the photogrammetric processes and 
  
e functions that are performed by the conventional photogrammetric equipment. 
n It is worth noting that some of the functions and processes that can be 
y simply and efficiently carried out for instance on analytical instruments are 
ie quite difficult to implement in a digital system. Even the performance of 
Tr simple elementary processes such as the image rotation with scale change, the 
m tracing and measuring in dynamic mode at acceptable roaming speeds and the 
d pictorial displays of stereo images require for example the resampling of 
n large arrays, the implementation of special algorithms and the special 
'S hardware support for fulfillment of conditions imposed by real-time 
S functions. The problems are even more complex on the higher levels of 
> merging the digital photogrammetric techniques for processing of metric 
le information with the techniques of digital image analysis where some of the 
le latter techniques (e.g. edge enhancement) may adversly influence the metric 
te quality of images. Consequently it is evident that the design of a universal 
S photogrammetric digital image processing system represents a difficult and 
il expensive long-term task. ; 
"9 
m DIGITAL SYSTEMS FOR PHOTOGRAMMETRY OF DYNAMIC PROCESSES 
f The technological advances in the development of solid state cameras with two 
dimensional arrays of detectors have contributed to the implementation of the 
concept of genuine or true real-time digital photogrammetric techniques. The 
  
LC main characteristics of solid state cameras (also called matrix or array 
n cameras) are: high speed of analog-to-digital conversion, direct digital 
m storage of images into memory at frame rates of 30Hz and higher, stability, 
1, wide spectral range, and small size. The detectors in the arrays may be 
m charge-coupled devices,  charge-injection devices or photo diodes. The 
ie digital image storage is the same two dimensional array of pixels (digital 
of gray level samples) as the one already mentioned in connection with scanning 
al and digitizing of photographs. On the basis of these characteristics it is 
e evident that solid state cameras (which are in a sense the digital 
1e counterparts of analog frame cameras) combined with appropriate high speed 
of computing devices and fast access memory devices represent the technological 
ne foundation for a new branch of photogrammetry that could be named "dynamic 
Le photogrammetry". 
al 
be Despite the low resolution of presently available solid state cameras 
1, (typically ^ 400x500 pixels and exceptionally ~ 1000 x 1000 pixels) the 
ag instrumentation build around them has already proven its capability for 
ic real-time measurement, monitoring and control of dynamic processes (Real and 
by Fujimoto, 1984). The meaning of real-time in this context implies the