2
Eleventh Congress of the
International Society for Photogramrnetrv
Lausanne, Switzerland, July 8-20, 1968
Invited paper for Commission III
Review of Strip and Block Adjustment
During the Period 1964-1967
G. H. Schut
Photogrammetric Research, Div. of Applied Physics,
National Research Council, Ottawa, Canada
Introduction
he period 1964-1967 is characterized by
the further development and the success
ful completion of a number of computer pro
grams for the simultaneous adjustment of
aerial photographs in large blocks. Both the
direct solution and the iterative solution of
the resulting system of normal equations have
proved to be entirely practical.
Block adjustments in which models, sec
tions or strips are adjusted as units are still
far more common. They can be divided into
two main groups: adjustment of models or
sections by means of similarity transforma
tions, and adjustment of strips or parts of
strips by means of polynomial transforma
tions of higher than the first degree.
The proponents of the adjustment of photo
graphs view this adjustment as the most
rigorous solution and as the main trend in
computational photogrammetry. Others re
gard the similarity transformation of models
as the true or rigorous least squares adjust
ment and the ultimate solution, and they
predict a decrease of interest in the poly
nomial adjustment. Nevertheless, the poly
nomial adjustment has many adherents and
is much used because it is the easiest to pro
gram and to use, and because it gives a very
satisfactory accuracy for topographic map
ping.
Any interest that may still exist in the
analog adjustment of input data has not re
sulted in more than one published paper.
Simultaneous Adjustment
of Photographs
1. USE OF THE COLLINEARITY CONDITION
References [5] to [23] deal with the simul
taneous adjustment of photographs in blocks
or strips. References [24] to [31] discuss the
solution of the large system of normal equa
tions in this adjustment.
Already before the London Congress, H. H.
Schmid and D. C. Brown used the condition
of collinearity of image point, perspective
centre, and object point for the simultaneous
adjustment of a set of photographs. This
condition leads to the linearized equation
v + Bb = t (1)
in which
v is the vector of corrections to the'photo-
graph coordinates,
5 is the vector of corrections to the
parameters of camera orientation and
to the coordinates of object points,
B is a matrix of coefficients, and
e is the vector of residuals of the photo
graph coordinates in the non-linear con
dition equations.
The normal equations become
-Pv + k =0
v + Bb = E
B l k = 0 (2a)
in which P is the weight matrix of the ob
served photograph coordinates and is com
puted as the inverse of the covariance matrix.
Further, k is the vector of Lagrange multi
pliers or correlates, and the superscript t indi
cates the transpose. Elimination of v gives
P l k + Bb = e
B‘k = 0. (2b)
Subsequently, elimination of k gives
B l PBb = B l Pz. (2c)