EZ SOIT 
  
  
It is shown from table 3 that the influence of 
lens distortion for three cameras are considera- 
ble.The camera "seagull" DF-1 135, consisting of 
six pieces of four groups of lenses, belongs to a 
good quality of amateur camera in China, but its 
distortion influence is 3 times bigger than that 
of MAMIYA of Japan and bigger than that of 
“seagull” 4D 120 too. If these cameras are 
immediately applied to photogrammetry such as 
stereophotogrammetry, the accuracy of determined 
object point is low from the computational results 
using the estimation formula of stereophotogramme- 
tric accuracy AY=Y2.mp /Bf and letting Y-10m, 
B=1.2m and m, =3r(table 4). Consequantly, the lens 
distortion must be calibrated and corrected in 
photogrammetry in order to expand the practical 
value of non-metric camera. 
Table 4 Stereophotogrammetric Accuracy Using 
the Non-metric Camera Immediately 
  
  
Model of Ay Y Relative 
camera (mm) (m) accuracy 
MAMIYA C330 144 10 1/69 
Seagull DF-1 422 10 1/24 
Seagull 4D 231 10 1/43 
  
2.3.2 Analysis of the average principal 
distance and lens distortion of non-metric camera: 
The average principal distance (f) and lens 
distortion coefficient (kl) for two kinds of 
camera can be obtained from table 1 and 2, the 
standard errors (m, and my, ) are estimated from 
the difference between average and each 
computational value, and the biggest displacement 
(5rm=Mu, (X-Xo)r2) due to the standard errors of 
distortion coefficient are derived. The results 
are in table 5. 
Table 5 Results After Correcting by Mean 
Distortion Coefficient 
  
Model of Eo iM Bp m(k1) rm 
camera (mm)  (:1075) (mm)  (:1075) (mm) 
  
"Seagull"DF-1 58.814 3.703 0.027 1.05 0.008 
"Seagull" 4D 75.853 .5619 0.028 .548 0.019 
  
The under knowledges could be found from above 
tables: 
(1) The changes of principal distance for two 
cameras are near. The standard error of changes 
are 0.027mm and 0.028mm respectively. The 
coordinate error on photograph due to the change 
of principal distance is 5x=0.007mm for camera 
135, and $xz0.008mm for 120, estimated by formula 
5xz(x/f)-mf. 
(2) The changes of distortion coefficient (k1) of 
film 135 in the process of photography for all 
film roll are small. Comparing with the mean 
distortion coefficient (k1), the standard error of 
distortion coefficient changes (mkl) is decreased 
by about 34 times. The biggest effect (6rm) of 
remainder distortion after correcting all film 
roll by mean distortion coefficient is only 
0.008mm. While the camera after correction of 
distortion is applied to stereophotogrammetry, the 
error of determined points AY equals 11.4mm and 
the relative accuracy can attain to 1/880 in 
photographic distance 10m and base 1.2m. 
(3) The changes of distortion coefficient for all 
film roll in camera 120 are greater than that in 
135. Comparing with the mean distortion 
coefficient, standard error of these changes is 
decreased by about 9 times. The biggest effect of 
remainder distortion after correcting all film 
roll by mean distortion coefficient is 0.019mm. In 
same case, the error of determind points AY equals 
21.1mm and the relative accuracy achieves to 1/470 
while the camera after correction of distortion is 
used for stereophotogrammetry. 
(4) Generally speaking, it would seem that the 
effect of the remainder distortion in camera with 
greater lens distortion after correction is higher 
than that in camera with smaller lens distortion. 
But from our results,it is the contrary situation 
that the accuracy of camera 135 after correction 
is better than 120. The average residual of image 
points obtained by residual of resection (table 1 
and 2) is 0.007mm for camera 135 and 0.014mm for 
120. The accuracy of the latter is lower than that 
of the former by half. We suppose the reason may 
be that the film 120 contains the bigger unequal 
deformation. The film deformation is caused by the 
expansion and contraction and unflatness 
influences. The area of film 120 is larger than 
that of 135 as stated above. Because there is not 
the flattening device in non-metric camera, the 
area suspended in air of film 120 in photography 
is greater than that of 135 by three times 
(John,1988),and the large areal unequality could 
cause the large unequal deformation. Previously, 
the equal deformation of film is only analyzed 
with the change of frame span in middle part of 
photograph and the simple ratio correction is only 
added in computation. It appears that for camera 
120, the unequal deformation of film has to be 
considered further, and corrected by the affine 
transformation using four or even eight 
collimation marks. 
3 CONCLUSION 
~The accuracy of non-metric camera made in China 
can be improved considerably by means of the prior 
calibration to determine and correct the average 
lens distortion and the film deformation. Thus by 
this method, the non-metric camera could be 
immediately applied to some photogrammetries with 
middle and low accuracy, like the metric camera 
with lens or without control points. Comparing 
with calibration in operation, the way of the 
prior calibration and correction could even more 
contribute to the popular application of non- 
metric camera. 
The lightweight stereocamera developed by our way 
has the characters of the uncomplicated structure, 
the convenient utilization and the more stable 
relative orientation (Wang,1988). It can be used 
in measurements of the geology, the traffic 
accident, the criminal scene, the simple and rapid 
architecture and the small areal topography as 
well as the seismology, having the wide prospect 
of application to the close range photogrammetry. 
Reference 
(1) John,Fryer,1988,Lens Distortion and Film 
Plattening: Their Effect on Small Farmal 
Photogrammetry. In: Int. Arch. phtogramm. and 
Remote sensing, Kyoto-Japan,Vol.27,part B5, 
pp.213-222. 
(2) Wang Li,Wang Qingting and Fan Qiulin, 1988, 
The Lightweight Stereocamera and Its Application 
to Seismology and Geology, In: Int. Arch. 
photogramm. and Remote Sensing, Kyoto-Japan, 
Vol.27, Part B5, pp. 365-372.