Comm. II, I.S.P.: Preliminary report
on standardization of expressions for
accuracy in photogrammetry. App. 2.
Determination of the accuracy of a stereocomparator.
l. Measurements of single image coordinates.
The instrument is tested with the aid of a glass grid of high and known
precision. The grid is adjusted in the comparator so that the coordinate
directions of the comparator approximately coincide with the coordinate
directions of the grid and so that the centre of the grid approximately coin-
cides with the origin of the image coordinate system.
The coordinates of a great number of regularly located grid points are mea-
sured in the coordinate system of the comparator x, y and are compared
with the corresponding known coordinates x/y'of the grid. Each coordinate
x and y is determined as the average of a certain number, for instance three,
repeated settings.
The errors of the measured coordinates x and y are defined as
dx = x'- x^ (1)
dy = y Ly (2)
These errors are assumed to be caused by regular (systematic) as well
as by irregular (accidental or random) errors.
Here we will determine primarily the following regular errors
a) Translation errors dx. and dy, of the grid
b) Rotation error do ofthe grid
c) Scale errors m. and m respectively of the comparator (affine deforma-
tions of the measured coordinates)
d) Lack of orthogonality between the coordinate directions of the compara-
tor. The angle between the positive x- and y-coordinate directions is
assumed to be 1008 + dB. See Fig. 1.
l.l Error equations.
Since the errors are assumed to be small, linear differential formulae are
sufficient to express the relations between the coordinate errors dx and dy
on the one hand and the regular errors dx . dy, ca , dm_, dm, and df
on the other.