ance with the reference model. The position of reference model
for this experiment is shown in Figele
2) ~ 32 mm
d.
20 — 32 mm
[:]
[-]
[]
eee
3 0 E17 4-1
1:
Fig. 2
The simplist method to be used is to regard the whole block
as an area with constant systematic errors, all models in the
whole block are superimposed on an unified reference model,
situated within a small square with tie points as centers and
sides length of 20mm, mean values for residuals of tie point
coordinates are computed, such values with reversed sisns will
act as corrections ,r systematic errors for the tie point.
2. Statistical check on corrections for systematic errors
at tie point
After obtaining corrections for systematic errors at tie
point, a statistical check has to be made in order to con-
firm whether significant systematicness exists or not, here
T-testing was used, its mathematical formulae are /1/ /5/
rx (ys /A/
where K refers to the number of sub-blocks
ng is the number of model points within the ith sub-
block
IA is the absolute value for the mean value of the
residuals within the ith sub-block -
RMSE is the root mean square error of residuals of photo
coordinates in blocks
When T <1 (corresponding to significance level (0.15,
n => 60), E(Ai)=O0, that is, no siznificant systematic error
exists, hence it is not necessary to make a posteriori |
- 491 -