ed.
(9
ur
nage
i
PU SE N MER NOBODY Sea DS quA ET Das
For image quality criteria in photogrammetry the most widely-
used method is the classical resolving power test. For locating
the plane of best average definition, the area-weighted average
resolution (AWAR) is commonly applied (1). It is obtained by
multiplying the resolution at different field angles by a
weighting factor proportional to the Size of the correspon-
ding image zone. Very frequently the geometric mean of the
tangential and sagittal (radial) resolving power is used.
Ideally, the area of the annular zones should be very small
so that the AWAR can be expressed as a limit. In practice, the
value of AWAR i's computed for zones of finite areas.
For obtaining the resolving power a test chart is imaged on an
emulsion. The analysis is carried out after development. Alter-
natively the resolving power can be found by the intersection
of the MTF with the threshold curve. The threshold contrast
curve takes into account the MTF of the emulsion as well as the
threshold contrast of the eye (11). Fig.1a) and b) show measured
MTF curves of a newer f/4 WILD aerial camera lens, focal lens
210 mm (WILD 21 NAgII) together with the threshold curve for
Agfapan 25 emulsion. In fig.2 the resolving power for this lens
is plotted as a function of the field angle (image height)
obtained:
- by classical imaging of a test chart and
- from the intersection of the MTF with the threshold curve.
Good agreement was found for the resolution with the two
techniques. Furthermore, the AWAR of the lens was 52 Lp/mm.
Image evaluation techniques based on MTF
Image quality assessment of aerial survey lenses on the basis of
resolution tests has some marked disadvantages; in particular,
resolution can only be measured for a given object contrast. In
recent years, the use of MTF for image evaluation has received
considerable attention. MTF curves should permit the photogramme-
trist to obtain a more reliable estimate of the performance of
the lens.
Other image evaluation techniques such as those based on edge
gradients and spread functions will not be discussed here.