N = D / 2K0 (3.3 )
where o0 1s the standard deviation in measured density for a number of
density levels, 0 the density range of the emulsion and K the number
of standard deviations by which adjacent density levels should be
separated. Thus K is a quantity related to the confidence of
distinguishing between adjacent levels. 60 = 2Ko will correspond to AD
i formula ( 3.4 ) and we get:
Mr x 6D / Y (03.6)
Here 60 is no longer the minimum density difference required to re-
solve a target of a certain contrast ( modulation ) but the density
interval represented by one level according to ( 3.5 ). The evaluation
is made for a certain emulsion without considering the properties of
the target or the human observer, focusing on the effect of film
granularity.
4. NOISE IN THE PHOTOGRAPHIC IMAGE
The photographic image 1s degraded by noise. The main source is the
grain noise. The grain in the photographic image varies in size and is
unevenly distributed in the emulsion. This causes variations in
density even if the imaged object is homogeneous.
The noise affects the digitizing in two ways. It is the limiting fac-
tor of both spatial resolution and grey level resolution.
4.1 Granularity
Granularity is an objective measure of the grain noise in the image.
The grain noise sets a limit to resolution and to modulation and also
influences the quantization. Granularity is usally expressed in den-
sity units and in terms of the root mean square (rms) value of the
statistical fluctuations in density, o.. The measurements are made in
a microdensitometer with an aperture of 48 um at a density level D=z1.
These conditions are used for example by Kodak when they supply
granularity data for their emulsions. In the data sheets the measured
O0 1s multiplied by 1000.
Investigations have been .made on the relation between granularity and
the area A of the aperture of the measuring device. As a result it was
been found that o JA = constant is a valid relation (Billingsley,
1375). This makes it possible to convert granularity data supplied by
manufacturers to fit the aperture of the actual digitizer used.
Measurements of grain noise are generally expressed in terms of
density. The sensor in the digitizing system is sensitive to light,
‘which has a linear relation to film transmittance rather than density.
For this reason transmittance is the important parameter and noise
considerations should be in terms of transmittance noise rather than
density noise. Density noise on the other hand is important when we
consider the visual effects.
144