13.
obtained from triplet relative orientation are applied to the original
measured plate coordinates of images. Triplet coordinates calculated using
corrected plate coordinates should yield a single set of most probable
coordinates regardless of which equations 2.03, 2.05 and 2.01 are used.
3. Most Probable Coordinates by Least Squares Adjustment Using
Co-Linearity Condition Equations
This approach involves forming two co-linearity condition equations
for each of the corresponding rays. Redundant equations written for a
pair or triplet of rays are then solved in a least squares adjustment in
which random errors in plate coordinates are minimized. It should be
emphasized that in such an approach, the least squares adjustment applies
only to the rays involved for one point. In other words, the adjusted
orientation parameters are assumed fixed from triplet relative orientation.
Since only one point is considered per computation, the problem reduces to
the least squares adjustment of a point triangulated from multiple camera
3
stations. This method has been fully developed and is similar to an
2
approach taken by Brown .
All three of the methods have been tested. The final computer program
includes method 1 for initial approximations and method 3 for the final
coordinates. Photogrammetrists may find that any one of the methods has
advantages depending on characteristics of the work being performed.
C. TRIPLET ASSEMBLY
Triplet assembly is performed in the following manner: (a) relative
orientation is perforned for the initial triplet assuming the middle photograph
of this triplet as the origin of the arbitrary triplet coordinate system
(Figure 2.01); and (b) since the third photo of the initial triplet is the
middle photo of the second triplet...etc., relative orientation for subsequent
triplets takes place independently for each triplet but with the position and