Guerra, Francesco
2W: NEW TECHNOLOGIES FOR THE GEOREFERENCED VISUALISATION OF HISTORIC
CARTOGRAPHY
Francesco Guerra
Laboratorio di fotogrammetria - CIRCE
Istituto Universitario di Architettura di Venezia
S. Croce 1624 — 30135 Venezia
ITALY
guerra2(a)brezza.iuav.it
Working Group V/5
KEY WORDS: geometrical transformation, geo-referencing procedure, warping algorithms, cartography, software
ABSTRACT
The procedure of referencing-transformation, set up and used by the Photogrammetry Laboratory of the IUAV, is based
on the application of two types of sequential transformation; first, a global transformation and then local
transformations.
These two types of transformations are each carried out on the basis of two sets of corresponding points, the first control
points, which could originate from a survey, a map or an image, the second set identified on the image to modify.
While the same points could very well be used, the characteristics of the transformations are rather different since the
global transformation is controllable in its effects through parameters which model the transformation expressed in
analytical form, while the local transformations are controlled only through the use of the features which define them.
The global transformation can be chosen on the basis of its properties and the result is therefore predictable and
controllable. In the case of the example where one wishes to transform one map onto another, while maintaining the
original shape of the objects (for example, the buildings) it is important to choose a similar transformation.
The local transformations, either using the finished element technique or the technique using force fields (warping), do
not offer the opportunity of imposing geometric characteristics as a priority, but are based only on elements placed in
correspondence on the source image and the destination image.
The result obtained after the local transformations is the transformation of the original image according to a certain set
of features, defined by pairs of homologous points. The main characteristic of this image is to be adapted as much as
possible to its reference since between the local transformation algorithms, those leading to a superimposition of the
control features were the ones chosen. In this way, the attributes of the initial image are maintained but its geometry has
been changed.
If the initial image is a map and the reference is another map, what is obtained is the transfer of the semantic content of
the source map on the reference map. This is the same as saying that the source map has changed its metric content, or
that it has changed its geometry.
Essentially, with the procedure of referencing-restoration-transformation, the metric nature of the initial map is lost: the
map has been mapped on the reference map. The term “mapping” is exactly in the meaning given to it in computer
graphics: the adaptation of a texture to a form.
In practice, one can imagine that a map (or image) is dressed with a metric-geometric support which can be changed.
1 PROBLEMS WITH TRANSFORMATION
The main problem with this type of procedure is that giving a new geometric support to a map in order to assign it a
new metric nature (the mapping of one map onto another), often radically modifies the original map.
This is more evident when the images are different from each other, while still representing the same objects. In the
case of the photoplane of Venice, taken in 1911 (see the illustration), the problem did not arise since the images (the
maps) were very similar to each other and therefore, the deformations, introduced with the local transformations,
compared with the original cannot be determined.
Having identified foundation and fundamental points of the cartography, which are:
e the measurement of the sign,
e the position of the sign,
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 339