Automated extraction of beach
bathymetries from video images
Laura Uunk MSc Thesis
prof. dr. S.J.M.H. Hulscher dr. K.M.Wijnberg
Contents
Beach bathymetries by shoreline mapping Manually mapping shorelines (IBM)
Automatically mapping shorelines (ASM) Problems encountered
Automated quality control
Automatically vs. manually obtained bathymetries Beach behaviour
3
Beach bathymetries by shoreline mapping
Argus images
• Time exposure images 10 minute average • Every half hour
Beach bathymetry mapped • Shoreline location
• Shoreline elevation • Throughout tidal cycle • Elevation data between low and high water
Manually mapping shorelines (IBM)
5
Manually mapping shorelines (IBM)
Requires many man-hours
•
up to 4 hours for one day for one station (5 cameras) Therefore no daily bathymetries, but monthlyOpportunities of Argus not completely used Automated version was developed (ASM)
•
PlantAutomatically mapping shorelines (ASM)
Human steps are automated
•
Definition of the region of interest> based on expected shoreline location on bench-mark
bathymetry
•
Quality control> compare detected points against bench-mark
bathymetry
7
Automatically mapping shorelines (ASM)
Database with shoreline points Shoreline points
within time window
Bench-mark bathymetry Shoreline elevation
Region of interest
Detected shoreline points
Accepted shoreline points Acceptance criterion
Detection method Elevation model start / next time step
Problems encountered
Bad bench-mark bathymetry
> bad definition ROI
> bad quality control Start of a downward spiral
9
Problems encountered - downward spiral
Database with shoreline points Shoreline points
within time window
Bench-mark bathymetry Shoreline elevation
Region of interest
Detected shoreline points
Accepted shoreline points Acceptance criterion
Detection method Elevation model start / next time step
Problems encountered - solutions
Better definition of the Region of Interest
•
large smoothing scales loess interpolation> better expected shoreline location
•
extension to edge of image> inclusion of entire shoreline
•
avoid zigzagging11
Problems encountered - solutions
Better expected shoreline location larger smoothing scales
longer time window small smoothing scales
Problems encountered - solutions
Better definition of the Region of Interest
•
large smoothing scales loess interpolation> better expected shoreline location
•
extension to edge of image> inclusion of entire shoreline
•
avoid zigzagging13
Problems encountered - solutions
Double quality control
•
Two bench-mark bathymetries> 1: small smoothing scales, small time window
> 2: large smoothing scales, large time window
Shoreline points first compared to first bathymetry
Points that could not be checked are then compared to second bathymetry
15
Problems encountered - solutions
small smoothing scale more detail more gaps
large smoothing scale less detail less gaps
Automated quality control
Fixed vertical criterion: Zdif
•
Sometimes accept points that are wrong•
Sometimes reject points that are good17
Automated quality control
What value should be used?
ASM was run with three values for Zdif
•
0.10 m;•
0.25 m;•
0.50 mASM bathymetries compared to IBM bathymetries
•
Coastal State Indicators (CSIs)> Contours (-0.50 m NAP; 0 m NAP; 0.50 m NAP)
Automated vs. manual
0 m contour for May 7th to 12th 2006
19
Automated vs. manual
0.10 m 0.25 m 0.50 m continued 0.25 m
21
Conclusions
• Man-hours are saved by automatically mapping shorelines
• Results automated version (ASM) correspond well with results manual version (IBM)
• 0 m contour by ASM shows immediate response of the beach to changes in wave height
•
this was not visible with monthly IBM bathymetries• Opportunities provided by half-hourly Argus images can now be fully exploited
• ASM data could be used to e.g.
•
study storm impact•
study influence of nourishments23