completed projects - selected points were deducted from previ-
ously produced orthophotographs. For all these points, the em-
phasis was on the correct interpretation of historical and con-
temporary images.
Given that the initial embedded parameters of external orienta-
tion images were very approximate, it was necessary for easier
automatic image correlation to measure connecting-identical
points on each photo, both in all lines and between these lines.
During this measurement the ground control points were simul-
taneously identified. During the measurement, the emphasis
was on the follow-up RMSE of tie and ground control points.
After measuring the connecting and ground control points, the
calculation of elements of the exterior orientation of single im-
ages on the basis of established limits was carried out. The re-
sults of the calculation were exterior parameters of aerial imag-
es, root mean square error on the ground control points, the
resulting sigma and various textual and graphical outputs for
subsequent analysis of the valuation, including error messages.
The aim of the AAT analysis was to examine the mean values
of residual errors of measured connecting points, the number of
redundant measurements, mean squared errors of the ground
control points and calculated angular parameters of exterior ori-
entation. The inspection of the ground control points’ distribu-
tion in the block was also necessary as well as the number of
measured and generated connecting points on the individual im-
ages in strips and between strips by means of graphical software
tools. Part of this analysis was also the connection and assess-
ment of the subsequent triangulation blocks characteristics.
After the AAT calculation analysis the erroneous connecting
points and ground control points were remeasured. Further-
more, the new connecting points at critical spots in blocks were
measured, and if necessary, new ground control points were
added. Calculation of AAT was considered final, unless the re-
quired AAT characteristics were achieved. The achieved aver-
age mean errors in AAT from the individual triangulation
blocks are listed in the Table 2.
Table 2. The achieved average mean errors in AAT from the
individual triangulation blocks
The resulting AAT accuracy is significantly influenced by the
defined (unknown) parameters of used aerial surveying cameras
and used ground control points, including their layout.
For images orthorectification, it was necessary to carry out the
radiometric correction on individual images - common editing
of image defects caused by light aberrations of the aerial camera
lens, ground reflections, field illumination, sun at different an-
gles, images processing in different seasons and different flight
parameters. These radiometric corrections were performed by
dodging in places, where the dark places were lightened and the
light places darkened, so that the picture has balanced presenta-
tion of details in shadows and more light-exposed areas. Alt-
hough a number of steps of radiometric compensation were
conducted automatically, high personnel effort was devoted to
these operations.
For the desired area of digital image processing and orthorecti-
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B7, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
fication the DTM necessary to transform the image on the real
terrain was prepared. The elevation accuracy of digital terrain
model is due to the resulting orthophotomap expected to be in
resolution of 0.50 meters to 1.5 meters. The basic problem of
the used DTM was its topicality, because in many places did not
match the area state during the years of images acquisition. For
the necessary corrections of the terrain model (updates removal)
the original technological process was elected leading to the re-
quired accuracy of the final orthophotomaps.
After the orthorectification was performed, the visual inspection
of resketched images was carried out, by which the geometric
quality was controlled and on the basis of the control points
wrong places were marked as well as places with image defor-
mations. In these designated areas the digital terrain model was
corrected stereoscopically to the state valid on the date of imag-
es acquisition.
After the DTM correction a new redraw followed as well as the
control of corrected places in the manufactured orthophoto-
maps.
Selected parts of archival images were defined using seamline.
These parts were used in the final mosaic. These cutting lines
were selected according to the appropriate configuration of the
terrain so the transition between photos was the least distinct.
After creating all seamlines the individual files with cutting
lines were assembled, cutting lines were checked for coverage
of desired space. Control of topology was carried out whether
each image corresponds to a polygon with one centroid, and
whether the cutting line had no loose end.
After finishing the calculation of mosaicking visual control of
all map sheets was performed for coverage of the desired area
without visible errors in the image on the resulting mosaic.
Places with geometric imperfections were marked into a special
file. Attention was mainly focused on the places where the cut-
ting lines cross with communications, line buildings and built-
up areas in general. Furthermore, the routes of especially dis-
tinctive communications were checked (main roads, railways,
roads on mounds or in pits); routes of larger rivers or canals and
places with bridges were marked in case there were some visi-
ble deformations caused by redrawing technology and other dis-
torted places caused by the terrain complexity or imperfect
DTM. During the cleaning process of the orthophotomaps view,
the individual places on it were scanned and the retouching of
major dirt was performed (dust, hair).
If necessary, some places on the orthophotomap were densitely
aligned, so the total orthophotomap gave compact impression.
Appendix 1 shows a sample before and after the final treatment.
The final products of the orthorectification were the individual
map sheets of orthophotomaps with a resulting resolution
of 1 m for orthophotomaps from pre-war period and 0.50 m for
orthophotomaps from the post-war period, in the map sheets
layout 1:5000 in the S-JTSK coordinate system, compressed
TIFF format, size from 6 MB to 10 MB with the relevant TFW
file bearing the data of georeferencing.
The resulting positional accuracy of the orthophotomaps is di-
rectly dependent on the AAT accuracy and DTM accuracy. Po-
sitional accuracy was verified with existing map data by select-
ing identical points evenly distributed throughout the area on
both historical and contemporary ortophotomaps. Coordinates
were checked at these points and the mean error was deter-
