What's New in Teams Calling, Meetings and Devices March 2024
mcneills_igarss2011_penguins.pdf
1. Semi-automated penguin counting
from digital aerial photographs
S.J McNeill K Barton P Lyver D Pairman
Landcare Research New Zealand
2. Motivation
Understanding changes in penguin population is important,
as these can be used as indicators of anthropogenic and
foodweb eects
Aerial photography is used in the Ross Sea (Antarctica) to
capture a reliable count of Adélie nesting penguins
From 1981, the Ross Sea area (158 175
o o E) has been
surveyed annually
There are many diculties in achieving this census count:
Timing is critical,
Ground counting is dicult or impossible,
Counting using prints is dicult to control and validate.
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
3. Objectives
Determine if it is possible
to reliably detect Adélie
breeding penguins in
images
Generate software to
(semi-)automate the
census process.
Test, using an expert,
and optimise interactivity.
Pygoscelis adeliae)
Adult Adélie penguin (
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
4. Adélie penguins
The most abundant and widespread Antarctic penguin
10 million Adélie make up 80% of the Southern Ocean bird
biomass
38% of all Adélie penguins are found in the Ross Sea
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
5. Image capture
Images captured using a hand-held camera through the open
doors of a helicopter and/or C-130 Hercules
Hasselblad H1D with a Phase One digital camera back
Image size 5440 × 4080, 3-bands natural colour, TIFF
EXIF data provides date/time and aperture information
Typical ground resolution better than 0.5 m
Ten representative images were selected for analysis
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
7. Sub-scene example
870 × 510 sub-scene
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
8. Analysis
Human detection of breeding Adélie not straightforward
There are many similar-looking objects in the images
Proposed revised approach:
Detect the distinctive area of the colony
Only count penguins within colony area
Provide software features to easily add/delete penguins
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
10. Colony penguin detection
Background is largely monochromatic
Colony area covered in guano and has a red excess over
green or blue, with higher saturation
Use linear discriminant analysis to separate colony from
background, based on:
Natural colour counts (RGB) converted to hue, saturation,
lightness (HSL) space values,
Two-way interactions of HSL space values,
Aperture setting.
Classication followed by morphological opening and closing
dene the colony area
Penguins detected as dark local minima within colony area
Penguin objects pruned to upper threshold of circularity
P 2 / (4πA) to remove long thin objects
Adopt the centroid of the surviving objects as penguins
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
16. Editing facilties
Detection procedure does not count all real penguins
False penguins counted
Non-breeding penguins within colony
Penguin shadows or spurious dark objects
True penguins missed
Breeding penguins outside colony
Penguins indistinct compared to surroundings
Editing facilities required:
Overlap between photographs requires group deletions
Add or delete individual penguins
Check that penguins are not double-counted
Record of editing steps maintained
Number of editing steps requires single-click operation
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
17. Implementation
Software written in Matlab 2010b, deployed with Matlab
compiler
Census results stored for each captured image in a small le
Deployed for testing phase to a penguin ecologist
Second development phase to x faults and improve
interactive response:
Reduce memory overhead for each counted penguin
Reduce keystroke eort for additions/deletions
Add ability to count penguins within non-guano stained area
No problems reported after second phase deployment
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
20. Colony classication rates
Accurate colony delineation is very important
Requirement is for high true positive, low false negative rates
About 5% of images give poor results:
Due to very poor colony/background distinction
No clear reason for this poor result
CF001669 CF001720
IGARSS-2011, 25-29 July 2011, Vancouver, Canada
21. Conclusions
Semi-automated penguin counting is a pragmatic approach
Laborious counting automated; ne editing left for an expert
Software allows editing, maintains counts, stores results
Emphasis is interactive productivity
Acknowledgements
Ministry for Science and Innovation (funding).
Antarctica New Zealand (funding and logistics).
Helicopters New Zealand (ying).
Squadron 40, Royal New Zealand Air Force (ying).
IGARSS-2011, 25-29 July 2011, Vancouver, Canada