<CoAwareness>
<interpreter>192.168.2.1</interpreter>
<TimeStamp>
<when>2012-01-01T07:12:23Z</when>
</TimeStamp>
<Placemark>
<name>Hongkou Airport</name>
<styleUrl>#msn_ylw-pushpin</styleUrl>
<Point>
<extrude>1</extrude>
<altitudeMode>relativeToGround</altitudeMode>
<coordinates>121.8022611111111,31.1498805555556</coordinates>
</Point>
</Placemark>
<rsImage>test.tif<rsImage>
<GeometricCorrection>
10,10,121.800,31.1498
32,46,121.821,31.1497
50,165,121.834,31.151
</GeometricCorrection>
</CoAwareness>
Figure 5. KML description of Collaboration awareness
3.2 Collaboration Related Service
In order to support collaboration interpretation, three
collaboration related services were developed. They are launch
service, perceiving service and communication service. All
services are deployed on collaboration server. Launch service
supports to create an image collaboration interpretation. After
selecting images and interpreters, user invokes launch service to
upload the information in the form of KML/KMZ file.
Collaboration sever create a collaboration folder in response to
launch service. A collaboration folder includes one task
subfolder and several interpreter subfolders. In the task
subfolder, KML/KMZ files describing task images and
interpreters information are stored. In addition, task images go
through some processing operations, such as projection
transformation, background elimination, and image pyramid
generation. The interpreter subfolders store KML/KMZ files
describing interpretation results for interpreters respectively.
After collaboration folder was created, collaboration server
prepares task images for each interpreter. All clients have a
unified interpretation interface. Perceiving service supports to
share interpretation results in KML files among interpreters. An
interpreter can either share his own interpretation results or
obtain others’. If sharing results, an interpreter invokes
perceiving service to upload KML files. Collaboration server
stores the KML files to some or all interpreter subfolders and
configures the results to clients for visualization. If obtaining
results, an interpreter invokes perceiving service to request
KML files from specified interpreters. Collaboration server
requests for interpretation results by sends messages to specified
interpreters and transmits results to the requestor.
Communication service supports to provide written
communication among interpreters using Socket messages. On
the interpretation interface, there is an interpreters list. An
interpreter can select one or more co-operators for
communication in a chat window. All messages are conveyed to
interpreters via collaboration server.
4. EXPERIMENTS AND DISCUSSION
In order to show the availability of the proposed method, this
article performs some experiments of collaborative image
interpretation on GeoGlobe. The GeoGlobe is an extensible and
flexible geospatial platform for managing and visualizing
massive geospatial information. It can provide distributed users
with quick browsing and transmission of geospatial data. Here
this article shows an example of collaboratively identify
LIESMARS (State Key Laboratory of Information Engineering
in Surveying, Mapping and Remote Sensing) of China from a
multi-spectral image. Three collaboration clients (called A, B
and C) are employed to complete interpretation task. Client A
selected task image and launched the collaboration
interpretation. Client B identified the position of LIESMARS
using cyan star icon and annotation. The annotation states
interpretation result and corresponding interpreter. Moreover a
detailed introduction of LIESMARS will be given by clicking
the icon. Client C drew the coverage of LIESMARS using
yellow polygon and annotation. Results of both client B and C
are updated to Client A. As a sponsor, client A discusses with
other clients and votes for final results. Then final results are
saved as KML/KMZ files. Figure 6 is a screenshot of
interpretation results.
Figure 6. A screenshot of the interpretation result
It illustrates that KML is available for collaborative image
interpretation. In addition, this method alleviates some
drawbacks of existed methods for collaborative image
interpretation. For image transfer, existed methods need transfer
task image file to all clients, which results in a high burden to
the network. KML-based method transfers image tiles instead of
image file. An image file may be more than a few MB or GB,
and an image tile may be a few KB. An image file only divides
into several tiles. Therefore, KML-based method brings a low
burden of the network. For collaboration awareness, existed
methods use either specified messages or application sharing.
Specified messages are hard for expansion. And application
sharing spends on lots of computing and network resource.
However, KML is convenient to be expanded for personalized
application. For example, users can define new geometry
element to describe complex plotting symbols. It provides better
performance on collaboration awareness. Moreover, existed
method saves interpretation results in image. If sharing results,
it needs to transfer image file. But for the proposed method, it
only shares KML files without image. Through the comparison,
KML-based method applied in collaborative image
interpretation has greater advantages than existed methods.
5. CONCLUSION
Existed methods for collaborative interpretation of remote
sensing images have many drawbacks, such as burden network,
inefficient sharing of interpretation results, slow browsing. This
article proposed a KML-based approach to improve these
problems. Experience results illustrates that KML-based method
provides better performance for collaborative image
interpretation. The KML has an advantage in describing various
spatial data. Also as an OGC standard, it is convenience to be
expanded and shared. It is possible that using KML to describe
and share interpretation results. Meanwhile, geo-browser can
210
pre
ge
rec
fir:
ser
ve
eff
it
int
an
Fo
“N
20
De
“Fi
Ch
De
inti
of]
Tas
coll
Int
Di
dist
geo
on
15,
IEE
517
Hu,
pro
Hu:
