The stereoscopic vision of the object permits the 
definition of a mesh of points with known 
coordinates. Such coordinates constitue a DTM, 
that is to say a not continuous description of the 
shape of the object. Obviously to a greater 
density of points in the DTM corresponds an higher 
accuracy of the recostruction of the digital image 
of the object, seen in its real position in the 
space (Fig.2). 
A greater density of points of the mesh must be 
accompanied, however, by a contemporary increase 
of the precision used in determining the 
coordinates of the points in the space. 
Infact it is useless to thicken the meshes of 
points belonging to the DTM and to reduce the 
dimensions of the pixels of the image projected if 
to these operation it is not associated a 
precision in their definition in the space, 
certainly smaller of the dimensions of the single 
pixel. 
The achievement of a dense mesh of points defining 
the DTM is strictly connected with systems of 
automatic determination of DTM itself. These 
methods have the good quality to define the 
spatial coordinates of a great number of points in 
a short time but imply a lot of problems in the 
use of photogrammetric takens with critical 
orientations and in the interpretation of the 
discontinuities present in the surfaces under 
study. 
In particular when the automatic software meets a 
sudden change of depth (break lines), it is not 
able to understand the real thickness without the 
intervention of an human operator, often not 
expected in these applications. It must be also 
considered that such systems of automatic 
definition of digital terrains are mostly for 
working in cartography, where the break lines and 
the sudden change of depth are only occasional 
(cliffes, precipices..) or present in some 
application (height of the eaves from the ground). 
At present in cartographic field interesting prog- 
ress in the projection of digital images of the 
terrain on a DTM have already been developed with 
the first experiences of the ETH in Zurich. 
Once defined the DTM, the shape of the object can 
be described with a sequence of triangular shaped 
micro-planes having as vertex three points of the 
DTM itself. 
Fondamentally the projection consists in the 
operation of projecting on the digital DTM the 
image of every pixel of the image, that means its 
tone of grey or color. 
    
Such a kind of projection, with the center of 
emanation in the front nodal point of the camera 
and the image placed in the space according to the 
angular parameters of the external orientation, 
can create a lot of problems of discontinuity and 
deformation of the image. For these reason a 
different method, called ‘inverse projection’ is 
preferred. Instead of writing the equation in the 
space of the line going through the front nodal 
point O (centre of the optic cone), the pixel P 
and intersecting the generic micro-plane S, it is 
preferred to write the equation of the line from 
the micro-planes to the nodal point. In this way 
it is granted the complete refilling of the 
micro-plane of the DTM even if it can't assure the 
formation of a biunique relationship between 
pixel-image and its visualisation on the DTM. 
Truly said the problem is not solved but avoided. 
In fact only in presence of a DTM so thick to be 
considered continuous it is possible to 
hypothesize a biunique relationship between the 
pixel-image and every pixel projected on the DTM. 
However it is important to remember that such a 
thickening of the mesh of the DTM is obstructed by 
important problems of calculation and by the 
precision with which the position of the points 
can be evaluated. 
As alternative to a thickening of the DTM, even if 
with different results, there is the possibility 
to subdivide every micro-plane in different 
micro-areas. From their barycentre it can be 
written the same equation of the line going 
through the front nodal point O too, used in the 
method of the ’inverse projection’. 
This method can help to bring the image projected 
on the DTM near to the real one, creating more 
pixel on the image projected in comparaison to the 
number of those ones belonging to the original 
digital photogram. In this way it is more granted 
the continuity of the image but we loose 
definitively the relation pixel to pixel. 
The time of computation and the study of the 
relationship between density of the DTM - number 
and dimension of the pixel of the photogram - 
scale of the photogram - scale of the image 
projected on the DTM are all together indicating a 
new direction for the present research. Besides, 
the solution of these problems is also strictly 
correlated to the development of new instruments 
of digital image and computation treatment. 
The acquisition of the images with classic 
photogrammetric cameras can be supported by an 
acquisition with CCD cameras directly in digital 
form. In particular the interest is toward the new