rs
repre-
or the
atching
photo-
proce-
7 finite
luction
nuities.
ling ir-
subse-
he cor-
ntegra-
cedure
ments.
arame-
ion. If
' block
lation.
instru-
ere the
R15. À
gy has
orrela-
dently
Subse-
trans-
In this
ole for
iat the
plotter
verifi-
digital
the ut-
g. It is
an aer-
adable
ossible
results
estab-
yraphs.
oncy of
proven
methods for photogrammetric plotting. The eye is able
to distinguish only about 10 grey values, whereas a
CCD-camera supplies in general 256 grey levels or 8
bits. This direct comparison of the sensitivity of the
eye and the number of digits of a CCD-camera is how-
ever very biased, as the eye has a great relative sensi-
tivity whereas the sensitivity range for the analog to
digital transfer is predefined. That means that the
human observer will also be able to distinguish con-
siderable texture, even in an area of poor contrast,
whereas the digital camera with its 256 grey levels
might already be too insensitive.
If one tries to quantify the radiometric requirements of
a digital camera, then it is useful to establish a relation
with the photogrammetric characteristics of an aerial
photograph. The density range of a black-and-white
photograph can be estimated at between 0.1 and 1.5,
possibly 2.0 D, whereas color photographs can have a
density range from 0.2 to 2.5 D. According to our own
investigations, one can state that black-and-white pho-
tographs with poor contrast still allow a human ob-
server to distinguish density differences of 0.01 D. Sim-
ilar figures are obtained with the Weber-Fechner ex-
periment (cf. for example [1]). This sensitivity to the
contrast tends to be consistent for a human observer
over a large density range and decreases only in
strongly blinding or very dark regions.
The sensitivity of a digital cameras is usually propor-
tional to the brightness. If one refers this sensitivity to
photometric quantities, one should take account of the
transparency. Fig. 1 shows the effective measuring
values of a digital camera. One can see that the discret-
ization of 256 grey levels gives a very inhomogeneous
resolution; whereas bright areas have a very discrimi-
nating definition, rather large steps are obtained in
Grey values
A
300 —
250
200
150
100
50
| lus hoá
80 100
Transparency (%)
Fig. 1
darker areas. The radiometric resolution being greatly
limited by the noise, the noise of the electronic device
as well as the noise of the photographic image should
be taken into consideration. Reasonably, the noise of
the camera should remain beneath the noise of the
image. Even the noise of the emulsion is not invari-
ant and can be largely reduced by an appropriate illu-
mination. For example, the noise of the grain of the
emulsion is largely amplified by directed light, where-
as a surface illumination can largely reduce this effect.
2.1 Image digitalization with the analytical plotter
DSR15
For image digitizing, the Institute of Photogrammetry
essentially uses the analytical plotter DRS15. Test mea-
surements have also been made on a Perkin-Elmer
scanner, on a Hell scanner and on the scanners of In-
tergraph and Vexel. Due to its flexibility and to the
high image quality obtained, preference was given to
the analytical plotter DSR15, equipped with a CCD-
camera (image matrix 512x512 pixels, image resolution
8x12 microns). The relatively small image sections
scanned by frames of only 4x5 mm require the assem-
bly of the images. Furthermore, overlapping surfaces
are meaned out.
In order to assure a rather homogeneous sensitivity
over the whole image density, the image is scanned
with different brightnesses (up to 5x). The camera
allows such a procedure as a blinding of neighbouring
pixels is avoided by the electronics of the camera. Fur-
thermore, the illumination has been modified by the
introduction of a diffuser in order to obtain incoherent
light. The mounting of a diffusing sphere would cer-
tainly have been more useful, but was not realized for
technical reasons.
Grey values
A
300 —
250 —3j
200 —
150. —
100.
so.
2
-
0 0.5 1 1.5 2 2.5
Optical density D
Sensitivity of the Philips CCD-camera presented in density and transparency. One notices the linear relation bet-
ween the measured grey density values and the transparency. As far as the density of aerial photographs is con-
cerned, one notices a strong decreasing of the sensitivity in dark regions.