 |
| Titel |
A novel approach for automated shoreline extraction from remote sensing images using low level programming |
| VerfasserIn |
Anastasios Rigos, Aristidis Vaiopoulos, George Skianis, George Tsekouras, Panos Drakopoulos |
| Konferenz |
EGU General Assembly 2015
|
| Medientyp |
Artikel
|
| Sprache |
Englisch
|
| Digitales Dokument |
PDF |
| Erschienen |
In: GRA - Volume 17 (2015) |
| Datensatznummer |
250107692
|
| Publikation (Nr.) |
 EGU/EGU2015-7404.pdf |
|
|
|
|
|
| Zusammenfassung |
| Tracking coastline changes is a crucial task in the context of coastal management and
synoptic remotely sensed data has become an essential tool for this purpose. In this work, and
within the framework of BeachTour project, we introduce a new method for shoreline
extraction from high resolution satellite images. It was applied on two images taken
by the WorldView-2 satellite (7 channels, 2m resolution) during July 2011 and
August 2014. The location is the well-known tourist destination of Laganas beach
spanning 5 km along the southern part of Zakynthos Island, Greece. The atmospheric
correction was performed with the ENVI FLAASH procedure and the final images were
validated against hyperspectral field measurements. Using three channels (CH2=blue,
CH3=green and CH7=near infrared) the Modified Redness Index image was calculated
according to: MRI=(CH7)2/[CH2x(CH3)3]. MRI has the property that its value keeps
increasing as the water becomes shallower. This is followed by an abrupt reduction trend
at the location of the wet sand up to the point where the dry shore face begins.
After that it remains low-valued throughout the beach zone. Images based on this
index were used for the shoreline extraction process that included the following
steps:
a) On the MRI based image, only an area near the shoreline was kept (this process is
known as image masking).
b) On the masked image the Canny edge detector operator was applied.
c) Of all edges discovered on step (b) only the biggest was kept.
d) If the line revealed on step (c) was unacceptable, i.e. not defining the shoreline or
defining only part of it, then either more than one areas on step (c) were kept or on
the MRI image the pixel values were bound in a particular interval [Blow, Bhigh]
and only the ones belonging in this interval were kept. Then, steps (a)-(d) were
repeated.
Using this method, which is still under development, we were able to extract the shoreline
position and reveal its changes during the 3-year period. Although simple, with minimal
human interaction and low level programming, this method can provide precise coastlines
with just one pixel resolution. Images covering more locations are currently under
consideration. |
|
|
|
|
|
|