22
is then processed by the computer.
Theoretically, the digital images to be processed are static, mo
nochromatic and graphical, so that the intensityF can only assume
two values:
f,
F(x,y)= ;
f
1
Actually, the MTF functions of the optical components: employed
in the whole process of image formation and acquisition (projec
tor, metric camera, photographic support, digitization device)
will degrade the quality of the signal to be acquired, in such a
way that the digitized data are defined by a multi-value discre-
te function.
where X,Y are the plate coordinates
In order to restore a two-values function which allows an automa
tic reading of the plate-coordinates, a set of preprocessing pro
cedures must be executed on raw data.
The sequence of the procedures is the following:
P1: detection of all parameters necessary to correct photometric
and geometric distortions;
Di: ,data correction;
P3: filtering;
P4: detection of the thresholding value, in order to obtain a
two-value matrix;
P5: thinning of the lines contained in the two-value matrix.
Besides, data relative to the reticle reflected by the examined
object cannot be generally extracted from one single photogram;
in fact, the informations about the reticis (that is, the signal
to be detected) are mixed with the informations about the object
itself (which has to be considered as noise).
In order to solve this problem, a pair of photograms have to be
taken under the same conditions; the one, with the object alone,
the other one with the reticle projected onto the object; the in
formations about the reticle are then obtained by subtracting
the pair of digital images. Consequently, the following procedu-
re must be performed:
PO: registration and subtraction of the two matrices.
In the following the resources and the procedures used to analy-
ze a sample image are described.
4. The Processing Structure
Vk
The architecture of the process structure is shown in fig.4.
The structure is bas on three interconnect
(D
Cu
ed system. (2)
fp
