1-
)"
ly
on
of
le
nt
of
nt
id
y
le
terpretation of characters that identify the previous Land Cover.
Visual identification was performed with digital images on the
screen. Identified was:
1. Pushing the borders of individual Land Cover
categories
2. Areas with different Land Cover
3. Lines and distinctive shapes which differ from the
surrounding area indicating different usage.
Identified differences were compared with historical source ma-
terials The historical source materials are:
1. Memories of witnesses and field survey
2. Maps from IL military mapping from the years
1836 - 1852
3. Maps from III. military mapping from the years 1877
- 1880
4. Cadastral maps since 1928
5. Topographic maps since 1952
6. aerial photographs from the years 1953, 1968, 1976,
1984 and 1991
7. Historical orthophotomap from 1953. This orthophoto
was always used as first one, because the most
significant changes in Land Cover occurred after
1953.
Historical orthophotomap from the year 1953 is the most im-
portant material for the study of historical Land Cover of the
Czech Republic. For purposes of creation of this orthophotomap
aerial photographs and reconnaissance photographs from the
aerial photographs archive of the Czech Republic (managed by
Military Geographic and Hydrometeorlogic Office - MGHO)
were used. When selecting suitable aerial photographs from the
post-war period, the date of acquisition as close as possible be-
fore the time of collectivization during the years 1952 and 1953
was primarily respected. Verification of archives of aerial pho-
tographs proved that photography in the years after World
War II was relatively fragmented and cover of the entire territo-
ry of the Czech Republic was attained between 1946 and 1959.
Some places were not covered by the post-war images at all
(before the collectivization). These places were covered with
aerial photos from pre-war period and images taken in 1959.
One place was first covered with images in 1996! The Table 1.
clearly shows why the project was called the Historical ortho-
photomap the 1953.
ear Number Percen- Year Number Percen-
ofimages| tage of images tage
1937 48 0.24 1955 1 040 5.12
1938 29 0.14 1956 1 460 7.19
1946 61 0.30 1957 307 1.51
1947 239 1.18 1958 445 2.19
1948 69 0.34 1959 394 1.94
1949 701 3.45 1962 17 0.08
1950] 1711 8.42 1964 2 0.01
1951 712 3.50 1966 7 0.03
19821 2 111 10.39 1969 4 0.02
1953| 8093 39.83 1970 2 0.01
1954| 2863 14.09 1996 2 0.01
Table 1. The number of images from individual years
From the mentioned documents it is clear that aerial photos, es-
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
pecially in the western area - west of the city of Plzen, were
probably problematic from military and political view and the
aerial photography, here was not completed until 1959. From
the post-war period a total of 20 240 images were selected and
from the pre-war period another 77 aerial photographs had to be
added. In total, 20 317 aerial photographs were processed. The
procedure of creating the Historical orthophotomap started with
selecting and production of the derived imaging basis from aer-
ial surveying archival images. Most of the images are
in18x18 cm size and in the scale around 1:25 000. For some ar-
eas, especially the border areas, were used aerial photos in size
30x30 cm. The quality of the original images was significantly
different. The reason was using these images for mapping pro-
duction during these years, followed by transfer of flammable
celluloid base on the PET base during the nineteen eighties.
Aerial photos are variously scratched and partly dirty because
of previous handling. The data about the internal orientation
(calibration protocols) was not preserved by these images. The
procedure of preparation the derived imaging basis required a
considerable effort from MGHO. The required images were
first selected from the acquired images “atlas”in which images
are sorted by the year of photography. From the archives of im-
ages the negatives (or slides) were then physically chosen and
by copying with photo copiers with electronically controlled
contrast adjusting and subsequent black and white photo lab
processing, the duplicate aerial photos were created and conse-
quently scanned.
Scanning of the archival images basis was a challenging techno-
logical operation, which had a major impact on the quality
of the final orthophotomap. For this reason, before the scanning
of sub-blocks of images, the scanning tests were carried out
with an aim on finding the course of density curves characteris-
tics on selected negatives, and to set the optimal scanning pa-
rameters so that the resulting original digital images could be
easily manipulated in further stages of production. The demands
on standard cleanup of the original images from dust and me-
chanical dirt were followed; the cleanup was made with device
working on electrostatic basis; and with the principal of purity
of thrust glass and underlying glass of the scanner. After scan-
ning the image, the completeness of scanned image was
checked including frame markers. Geometric accuracy verified
by manufacturing plant calibration of the scanner was RMSExy
— 2 micrometers and digital image pixel size was 14 microme-
ters. After scanning all the images the country area was divided
into sub-blocks.
Measurement and analytical aerotriangulations (AAT) were
quite unusual. Based on the sheet index of historical photos the
flying strips and all aerial photographs were defined. The imag-
es were completed with information about the camera (specified
focal length parameters, coordinates of the fiducial marks, the
main point and lens distortion). Parameters of the cameras were
determined by autocalibration, because the calibration protocols
were not preserved (not even any partial information about the
elements of interior orientation, except for gross information
concerning the focal length of the camera) for the cameras used
for the acquisition of historical images. After the automatic
measurement of fiducial marks, images were measured manual-
ly in places, where the machine could not identify the fiducial
marks and in places where measurement error exceeded the lim-
it. Any corner fiducial marks could not be missed during these
corrective measurements. For each block the ground control
points were selected from the current documentation, in order to
be identical with points on historical images. Mostly it was the
church towers, intersections and landmarks in the field. Fur-
thermore, these points were supplied by selecting from other
