53 
This program controls the different function of 
stepper, as measuring mode, stepped by hand or automatically 
and so on. 
The optical structure of this plotter fol1 owes the 
idea of magnifying glass with telescope, when the magni f yi ng 
glass has to be a well corrected C photogr aphi c type} lens 
system in order to achieve acceptable optical resolutions. 
The optical enlargement may be changed by the focus of 
frontal lens system. In our case the enlargements are 7x, 1 Ox 
and 12x. 
A light measuring mark illuminated with special bulb has 
been built into the optical system, the illumination may be 
control 1 ed. 
The software of the Analpret are written in Borland 
Turbo Basic, which produces compiled machine code EXE files. 
The GRIDCOR serves to produce a correction array and 
software key for refining parallelity and perpendicularity 
within and between the plotter coordinate systems. The array 
and correction equations are used in the applied measuring 
programs. 
INTOR using the GRIDCOR constants produces the affine 
transformation parameters for calculating image coordinates 
from plotter coordinates and back. The program evaluates the 
measuring result by T test. The residuals are between 
0. 01 -0. 03 mm. 
RELABS uses the GRI EX} OR and I NTOR arrays and geodetic 
coordinates of control points for on line relative and 
absolute orientation. The program uses statistical test and 
gross error investigation for judging the quality of the 
whole measuring process. Using signalised control points 
0.02-0.04 mm RMS may be achieved related to the image plane. 
The coordinates of new points may be determined within 
the RELABS when the PY control automatically eliminates the 
Y parallax. The transformation parameters of RELABS are saved 
automatically. They serve for driving NEWPOI NTS which is 
planned for further application, for huge quantity of new 
points for a file . The Py control seems to be faster in 
NEWPOINTS as the program itself smaller than the RELABS. 
The automatic Py parallax removal may be considered 
"quasi" on line if the model is pointed continually. 
The idea of Py removal is demonstrated in Fig. 2. As it 
is known, there is no Y parallax in the normal case of 
orientation. The transformation of points with the relative 
orientation elements gives the key for this correction as it 
is explained in the figure. 
We have also dealt with the parallax removal in case of 
space intersection by col1inearity. It is observed that 
residuals in adjustment are loaded with gross errors caused 
by the y parallax. It may be logically accepted, that the 
residuals of observation equations, which keep to the 0 if 
the Py correction is also O, will content the spread 
gross error of Py must be corrected automatically. If we 
introduce the sum of absolute value of the residuals as Py 
correction the Py parallax may be eliminated. 
A more exact solution introduces the Py correction as unknown 
into the observation equations.