=3- 
Errors or Variations in Component MTF's 
  
The shaded area in Fig. 2 represents the range of on-axis MTF's for 
the lens systems of two Wild RC 8's (Universal Aviogon), a Zeiss RMK AR 
15/23 (Pleogon AR), and a Zeiss RMK A 30/23 (Topar A) as measured by 
Rosenbruch (1976), Martin (1976), and Welch and Halliday (1973) using 
different techniques. The extreme variations in the measured lens MTF's 
ranges from about 10 percent response at 10 cy/mm to 20 percent at 100 
cy/mm. Variations of this magnitude are reduced to an insignificant 
level when the Tens MTF's are cascaded with other component MTF's such 
as that for 20 um of image motion (Fig. 3), indicating that variations 
in individual system components must be rather large before their effects 
on image quality will be noticed. For example, the range of predicted 
system resolution values (as determined by method 3a in Table II) due to 
the variation in lens MTF's is 36-41 Tpr/mm and 41-47 lpr/mm for 1.6:1 
and 2:1 contrast targets recorded under static laboratory conditions on 
EK 2402 film (Fig. 2), and 28-30 and 30-33 lpr/mm when 20 um of image 
motion is introduced (Fig. 3). 
In actual tests of Wild RC8 and Zeiss RMK A 30/23 cameras loaded 
with EK 2402 film, laboratory resolution values for a 2:1 contrast target 
averaged 41 lpr/mm and operational values for a 1.8:1 target contrast 
(at the camera lens) averaged 27 lpr/mm, confirming the predicted esti- 
mates (Welch and Halliday, 1973). It is evident from this example that 
reasonable variations in lens (or film) MTF's (e.g. + 10%) are unlikely 
to significantly effect system performance as judged by MTF, resolution 
or overall image interpretability. 
Non-linearity of the Photographic Process 
  
The non-linearities of the photographic process introduced by the 
film D-log E curve and adjacency effects were discussed by De Belder, 
Jones, Sorem and Welander (1972) and remain of concern. However, non- 
linearity problems can be minimized by selecting targets that are recorded 
on the straight line portion of the D-log E curve and produce a AD of 
approximately 0.4 to 1.0 (Fig. 4a). The situations in Fig. 4b,c, repre- 
senting targets recorded on the toe and shoulder of the D-log E curve, 
Should be avoided. If MTF's must be derived from second-generation images 
(as in the case of Skylab) extreme target density values should lie on the 
linear portions of the D-log E curves of both first- and second-generation 
products. Assuming that targets are recorded on the straight line segment 
of the D-log E curve, the film y can be used to numerically compute the 
exposure values required for MTF calculations; thus avoiding time consuming 
point-by-point interpolations from the D-log E curve. 
Target Fidelity 
  
Most operational system performance evaluations are based on 
measurements of imaged edges (method 2a in Table II), and the size and 
sharpness of these edge targets is of critical importance. Size require- 
ments for imaged edge targets are determined primarily by the width of 
the system spread function which for most photogrammetric camera systems 
varies from 30-100 um, with 40-60 um representative of on-axis conditions 
for Wild or Zeiss Cameras employed with a typical mapping film. For an 
image recorded by a sensor system with a 40-60 um spread function a pattern