100. 00
+g
li
J
i
rd
i
for the first stage, with two iterations; P was computed for (x, y,
z) and X, Y, Z) respectively from:
EXP (-0,05(Kv/oo) **^ (2.4.4)
J
il
and à à
EXP(-0,05(V/00) ' ) (2.4.5)
ae
Il
for the second stage, with other two iterations; and recomputed from
EXP (-0,05(KV/00) ^*9 (2.4.6)
a
Il
and
FJ
il
EXP (-0,05(v/9o) ?*9 (2.4.7)
for the third and last stage, with two more iterations. K, in
equation (2.4.4) and (2.4.6) is a multiplication factor, that acts,
together with the initial weight difference (100 and 1) to show in
the down strip points coordinates, practically, real size residuals.
K was experimentaly established arround 4.7.
To test the algorithm, eight photo strips, analytically formed as
discrebed in 2.3, were used. Gross errors were itentionally
introduced in the point coordinates.
Table 2.4.1 gives the tie point coordinates in both strips. Table
2.4.2 presents the residual from least squares adjustment, and Table
2.4.3, the residuals when robust estimation was applied. Many other
practical tests were made, with the algorithm, that confirm the
detection power of the approach.
TABLE 2.4.1 Give the list of strips coordinates of tie points of
two adjacent strips (upper and down). Point 131 has
1mm blunder in everyone of its componentes.
UPPER STRIP DOWN STRIP
POINT X(mm) Y(mm) Z(mm) x (mm) y (mm) z (mm)
*131 1.979 87.058 -193.605 -0.011 -95.472 -189.027
132 95.136 86.305 -211.690 292.114 -96.922 -208. 397
125 141.593 132.142 -186.855 138.516 -50.993 -183.550
133 187.702 85.968 -161.114 184.832 -97.096 -158.024
134 278.679 84.636 -159.527 275.846 -98.159 -156.687
142 324.356 37.050 -203.477 321.554 -145.519 -200.855
135 370.730 83.099 -167.342 367.897 -99.411 -164.732
170 324.858 83.475 -182.109 321.977 -99.117 -179.396
* Point in which blunders were introduced.
- 449 -