a great economic impact in massive production.
The analytical transformation systems with analogue electronic image
transfer by means of CRTs represent another variant. Examples are
UNAMACE, Bunker & Ramo (Bertram, 1965) and GPS, Gestalt (Kelly,
1977). These systems imply processing of the video signals by
analogue electronic circuits.
Hybrid systems for image transformations have been bypassed by the
recent development of the digital systems.
Digital systems allow image transformations in geometric and
intensity/colour domains. Initially such systems were used for
transforming low-resolution remotely sensed digital images.
The geometric transformations are as in analytical systems. The
image window, however, can be represented by a square or rectangular
matrix of pixels, or even by a single pixel. If the corresponding
DTM is detailed enough, accurate geometric transformation is
possible for any type of terrain.
Digital systems have several potential advantages, as they can
process both geometric and intensity/colour information. Examples of
the latter are image enhancement and edge matching at automatic
mosaicking. Moreover, graphs and text can be easily overlaid on
images, e.g., to create image maps.
A transformed digital image can be composed of fragments from
different images of the same terrain. These fragments can be from
new and old input images (for updating), or they can represent
sections of different terrain slopes facing different camera
stations. Moreover, occluded areas in one image can be replaced by
the visible segments from the adjacent overlapping images, etc.
These potential advantages are impeded by the AD conversion of
images. Relatively large pixel sizes used at present (20 to 50 pm)
reduce the image quality substantially. The resolution of the
analogue images varies between 50 and 150 lp/mm, which corresponds
to pixel sizes between 2 and 7 pm. Hence, the amount of the required
digital image information would be 25 to 100 times more than at
present. We should increase the storage capacity, the speed of
storage/retrieval, communication, processing, and of the output
accordingly.
Image quality is further reduced by the resampling which is
associated with the geometric transformation of digital images. To
meet a specified ground resolution in the transformed digital images
with relatively large pixel sizes (at present), the scale of the
input images has to be increased accordingly. This, however, impairs
greatly the economics of production.