International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004
topographic maps (State Bureau of Technology Supervision,
1996). Each point in imagery is interpolated and positioned in
DEM strictly according to collinearity condition equation. Its
gray is given the gray value of corresponding point in DOM
based on the ground coordinate. For these simulated imageries,
their orientation parameters, image coordinates and ground
coordinates are known. It is very convenient to evaluate
precisely the practical accuracy of block adjustment. Table 1
shows the main technical data of each set of simulating
imageries.
Table 1. Technical data of aerial imageries
em Parameters
Imagery-1 Imagery-2 | Imagery-3 Imagery-4 | Imagery-5
strip number 11 10 5 4 iris
image number 11x15 10x12 | 5x7 4x6 Axa
observed point number per image 9-15 10-19 9-15 13-32 36-66 |
M. pomme eT ES 22424 moifio 10 62-10
.. . . Avemgeimagescae | 1:20000 140000 Gh ob |
padi ai > ame 23cmx23em " oo.
principal length 153.189mm mm d
“resolution of digital imagery a Te uk .
heading overlap 60% asie
. . side overlap. : : Tdi. | 4094 :
For conveniently adjusting, 9 standard orientation points are
extracted in each model of imagery-1 and imagery-3 and
transferred to the other imageries of différent scale. The
orientation points in imagery-2 and imagery-5 are all
transferred from imagery-1, and the orientation points in
imagery-4- are all transferred. from imagery-3. - If the-
transformation of orientation point at the standard position is
failure, it must be extracted again in the: current image to
confirm that the number of orientation points in a model is not.
less than 9 and there is at least 3 tie points between two
neighborhood models. During the experiment, random Gauss
noise with zero mean value is added all image observations to
simulate the measuring error.
Random noises produce by generator of random number. This
experiment adopt fake generator controlled by software. We
produce pseuso-random sequence that satisfies the requirement
of the experiment at one time. For instance, 4684 random
numbers are produced when imagery-1 is simulated. And they
are compiled with Gauss distribution. To make the series of
random numbers compiled with normal distribution, a series of
random numbers are firstly produced by multiplicative
congruential method and these numbers are complied with 0-1
distribution. Then these random numbers is standardized to
produce a series of random errors. The statistical result shows
that the mean value of the series of random errors is zero and
the standard variance is (1 0.01)x 5um . The series of
random errors are added to image coordinates of each
orientation point, so we get a set of image observations with
measuring errors of Sum.
Adjustment of imagery-l, imagery-3 and imagery-5 are
separately operated by conventional bundle block adjustment
software WuCAPSgps (Yan, 2000). A full ground control
points around the perimeter of the area at intervals of two
airbases and an elevation ground control points at intervals of
four airbases in the center of the arca are set up. Table 2 gives
the adjustment results.
The data in the table 2 shows that the m 'asuring accuracy of
36
image measurements is not larger than 5.7. m. It accords with
the standard variance of the added random errors in image
observations. The practical accuracy ‘of adjusted coordinates of
photogrammetric points is very close to the theoretic accuracy.
It also shows that the program of aerial simulated photography
is correct, and produced imageries can be used-im the following
research. A
4. EXPERIMENTS AND RESULT ANALYSIS
The exterior orientation elements of five sets of simulated
imageries and 3D coordinates of densification points are
calculated by combined bundle block adjustment with the
derived mathematical model. The adjustment methods are
divided into four ways:
A. Using imagery-1 to calculate the orientation
parameters of imagery-2 and the coordinates of
densification points;
D. Using imagery-3 to calculate the orientation
parameters of imagery-4 and the coordinates of
densification points;
C. Using imagery-1 to calculate the orientation
g gery
parameters of imagery-5 and the coordinates of
densification points;
D. Using imagery-5 to calculate the orientation
parameters of imagery-1 and the coordinates of
densification points.
None of GCPs is used in the above adjustment methods. These
bundle block adjustments are accomplished by combine the
image observations and the exterior orientation clements of the
fixed imageries. According to photogrammetric practice, the
exterior orientation elements adopt the ones calculated by the
conventional bundle block adjustment in the second section,
not the theoretic values of the simulated imageries. And image
observations of two periods of imageries have Sum measuring
errors. The adjustment results are shown in table 3.
Interna
Imager)
Note: 1).
2)
33.3
Met
- eo
From Tabl
1) Table
parar
grout
large
scale
unkn
betw.
bund
unkn
2) Ther
the c
two |
table
orien
pape
bund
3D €
betwi
