1e
a-
at
a-—
In fig.1 bank structures on the surface of à tidal flat area
are shown, The image shows a part of the tidal flats near
Hooksiel about 20 km northern Wilhelmshaven, Germany. Between
the banks there are depressions filled with water. These are
only a few cm deep. The distributions of the reflexion values
for the banks and the depressions, measured on the earth surface
do not overlap at that hour of the day when the aerial photo-
graph of fig. was taken.
This difference of reflection on the ground between banks and
depressions causes different greytone for the images of the banks
and the depressions on the IR-photograph. 9o it is possible to
Separate banks and depressions by image preprocessing.
So two types of elementary objects have to be distinguished,
the images of the banks and the images of the depressions,
which are complementary to each other,
Fig. 2 shows the amount of area of the banks and of both, banks and
depressions as measured from the image, The largest bank on ground has a
length of 20 m and a width between 6 and 10 m so that must have
a size of 6m x On x scale, This is a size of 5x5 pixel.
If the significance factor is set to 1, the perception value must
be 2/4, If the perception value would be smaller, the image struc-
ture would not be a texture in the whole image section. If the
image structure is not texture in the whole image section, if the
perception value is 2/4, then one might conclude that there is
a disturbance in the process of bank development,
Fig. ^ shows a part of the investigated image section. The image
was filtered before. In the image section shown in fig. 3 there
are no disturbances in the process of bank development.
This example shows that the choice of the perception value may
depend from measurements on the ground. For its computation in
this example mean and standard deviation of the common distribu-
tion of the amount of the areas of banks and depressions was
used, The small perception value shows a good homogeneity of texture.
If the image structure showing the banks and the depressions from
the surface of the tidal flat area is not texture, when the per-
ception value is equal 2/4 then there is a greater depression
between the banks or a greater bank between the depressions, so
the pattern of the banks respectively the pattern of the depres-
‘sions is disturbed. So the perception value can be used as an
arrangement parameter for one of the both types of elementary
object images.
6. Conclusion
For analysing textures three groups of parameters are important:
1. greytone variance parameters (as defined by Haralick)
2. shape parameters
5. arrange parameters
By greytone variance parameters a texture is not characterized
sufficiently. A correlation between parameters and properties
of objects on the earth surface is not possible.
The definition of texture is possible by using shape parameters
and arrangement parameters, These parameters correlate with pro-