1
|
i
I
THE SPATIAL ANBLOCK METHOD
Cornelis M.A. van den Hout
ITC, Enschede, The Netherlands
Commission III/3
The Anblock method, developed in 1960 [2], is a model triangulation
method restricted to planimetry only. This means that the models
must be pre-levelled to avoid disturbing the planimetric accuracy.
The procedure is non-iterative--the two-dimensional similarity
transformation being a linear problem.
The spatial Anblock method is an extension of the planimetric An-
block method. It starts to level (using equations) all models in a
block simultaneously.
Absolute and relative or independent ground control points (distan-
ces, slopes, lake points) are used to level the models with respect
to the object photographed. Relative ground control points, of
course, can be determined easier than their absolute coordinates.
Two levelling procedures will be described. The first is based on
distances and heights and the other on only slopes.
Finally, because the coefficient matrix of the partially reduced
normal equations of the Anblock approach has a very special struc-
ture, an efficient algorithm for the solution of the system of nor-
mal equations is described.
Bundle models.
A bundle model is used as a computational unit [11. The bundle model
is defined by the points of intersection of the projecting rays from
(measured) corresponding image points in a triplet of consecutive
and relatively oriented photographs and the perspective centre of
the central photograph. The relative orientation of a triplet is
determined by eleven independent elements of orientation. An effi-
cient relative orientation method was developed by Molenaar [4, Sh
The rotation elements are determined independent of the base ele-
ments.
If independent models are measured, bundle models are formed by
connecting pairs of consecutive independent models (spatial simila-
rity transformation) [1].
Bundle models are used because they have the same lateral overlap as
the photographs. This means that the number of the bundle models in
a strip is two less than the number of photographs. All points can
be taken into account in the adjustment, apart from the points in
the first part of the first and in the last part of the last model
in a strip. The perspective centres cannot be used to determine the
tilt of the models. They are used only if auxiliary data are avai-
lable, i.e., APR and/or statoscope measurements.