2004
| the
a has
1ages
| also
be a
r the
king
tion.
2 À
aken
>
72.
a are
ts of
Fer
'ency
data
earch
data
] put
] the
0gics
their
gling
y be
e-by-
n of
same
ation
y the
The
color
ation
ation
picts
d on
y the
axis
es in
nages
| this
nages
lution
od to
pping
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
Resolution
^
Figure 3: 3D representation of image resolution.
As an applicability example of the above visualization,
imagine a forester conducting a tree vitality evaluation. Our
visualization in figure 3 gives her an at-a-glance overview of
available resolutions within the spatial extent of her
interest.
For better navigation in the 3D model, the user can zoom
into an area of interest to further explore the available
resolution. This is done by a sliding window, which is
marked by an X in figure 3. The zoom-in version
overlapping the selected window area is shown in greater
detail in figure 4. This figure shows two images, with the left
image having a resolution of 2m and the right one a
resolution of 20m. To enhance the selection process the
original images can be superimposed on the 3D illustration.
The user can also rotate the 3D surface to investigate hidden
areas.
Resolution
5200 NO
14
Figure 4: Zoom-in of selected area from figure 3 with
superimposed images.
4.1.2 Multiple Instances of a Single Attribute: An
extensive data collection is likely to contain multiple
images of varying quality covering the same area. In the
following visualization (figure 5) the data with the highest
quality data value is chosen for the z-axis. Color is used to
convey whether there are additional datasets of lower
resolution available for the same area. The lighter (yellow)
color shows that there are no additional datasets, while the
darker (red) color shows that there are additional data (i.e.
high multiplicity).
Multiplicity
Resolution
Figure 5: Representation of additional available imagery in
lower resolutions (multiplicity).
Applying the above representation to our previous scenario
gives our forester access to additional information that she
would not be able to get solely from the 3D model. The user
would like to know what other datasets exist to facilitate
more detailed processes. This would be especially useful
when multiple images of different resolutions covering the
same area would be required. For example she might be
interested in a low resolution satellite image to get an
overview of forest density areas, and based on that,
subsequently use a high resolution aerial photograph to
extract the vitality of single trees.
4.2 Combination of Attributes
The next step is to combine information on multiple quality
attributes in one image. For this we use 3D representations
and superimposed color. Unlike the previous example that
used color and 3D to communicate different aspects of the
same attribute, below we discuss how to combine different
attributes in a single visualization.
4.2.1 Combination of Two Attributes: In order to
effectively combine the visualization of two attributes, the
first attribute is depicted along the z-axis of the 3D image.
The second attribute is conveyed by the color that is
overlaid. Figure 6 shows the combination of the attributes
of resolution and currency. The images of the highest
available resolution are displayed on the z-axis (as before),
while the currency is communicated with the help of
variations in color hue. Lighter shades of green represent
older data, while the more the color changes towards blue
(darker) the more recent the data is. When multiple images of
the same resolution but taken at different times are present,
we communicate the most recent available image.
This visualization supports users that are interested in the
currency of the data in addition to resolution. A city planner,
for example, whose task is to identify a site for a new
housing development wants to use geospatial images for a
first overview of the area. One of the important parameters he
considers in his evaluation is road access to the site. In order
to effectively do so, he can use the above visualization to
extract the most current data with sufficient enough
resolution to identify roads.
1005