Whose presence in the data would damage later steps.
2.1 Windowing of photocoordinates (S1)
This step only detects error whose estimated photocoodinates cause
the images to fall out of the limit of the negative.
2.2 Active Length invariant transformation of coorresponding images
(S2)
This step is useful specially to clean data from monoscopic
measurements affected by very large errors. The orthogonal model is
used to transform the coordinates of the image points from the right
to the left image of each stereopair. The transformation
X
X
x Cosk + y Sink + Xo
x Sink + y Cosk + Yo (2.2)
where (x,y) and (X,Y) are photocoordinates of the right and left
images respectively and K, Xo and Yo are parameters of the
transformation, causes the superposition of corresponding images,
neglected, the relief displacements and smaller effects. Gross
errors of greater magnitude than the effects of relief displacement
may be detected in x, but it is in y that it is possible the
detection of gross errors even of magnitude larger than the maximum
y .paralax. After applying the transformation, the corresponding
values are estimated and used, as the main tool of this step, to
detect and locate gross error. Table 2.1 shows the results of one
application of this approach. As it is seen from the table, and is
not surprising, the y discrepancy is the best detector and
identificator of blunders at this phase. Also gross errors from
wrong numbering of points and similar blunders can be successfully
detected at this phase.
TABLE 2.1 Showing at the first column the number of the points
common to both (left and right) images; 2nd and 3rd
columns presents the photocoordinates of the left photo,
4th and 5th columns, the transformed coordinates for the
same points and the two last, the discrepancies.
POINT PHOTO 1 TRANSFORMED DISCREPANCIES
X (mm) Y(mm) X(mm) Y (mm) DX(mm) DY (mm)
104 2.419 85.493 3.031 84.376 -0.612 1,147
105* 86.752 82.223... 92.923 85.692 -6.171 -3.469
112 47.477 44.287 41.441 43.701 6.036 0.586
119 0.901 - 0.656 Br 167222 0.507 -0.266 -0.149
120 90.520 = 2.239 85.493... — 22.933 9,027 0.694
127 41.499 -42.702 43.118 | -42.944 -1.619 0.242
134 - 0.643 -88.717 - 4.755 -88.170 4.113 -0.547
135 83.259 -85.966 83.104 -87.014 0.155 1.047
170 38.094 -78.410 44.755 -78.889 -6.661 0.479
* Point to which a. gross error |. of 5mm ..was itentionally
introduced.
- ja =