International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Volume XXXIX-B8, 2012
XXII ISPRS Congress, 25 August — 01 September 2012, Melbourne, Australia
LOW-COST, ULTRA-HIGH SPATIAL AND TEMPORAL RESOLUTION MAPPING OF
INTERTIDAL ROCK PLATFORMS
Mitch Bryson, Matthew Johnson-Roberson and Richard Murphy
Australian Centre for Field Robotics
The University of Sydney
Sydney, Australia
m.bryson@acfr.usyd.edu.au, mattjr@acfr.usyd.edu.au, richard.murphy @sydney.edu.au
http://www-personal.acfr.usyd.edu.au/m.bryson
KEY WORDS: Photogrammetry, Mapping, High resolution, Three-dimensional, Coast, Ecology
ABSTRACT:
Intertidal ecosystems have primarily been studied using field-based sampling; remote sensing offers the ability to collect data over large
areas in a snapshot of time which could compliment field-based sampling methods by extrapolating them into the wider spatial and
temporal context. Conventional remote sensing tools (such as satellite and aircraft imaging) provide data at relatively course, sub-meter
resolutions or with limited temporal resolutions and relatively high costs for small-scale environmental science and ecology studies. In
this paper, we describe a low-cost, kite-based imaging system and photogrammetric pipeline that was developed for constructing high-
resolution, 3D, photo-realistic terrain models of intertidal rocky shores. The processing pipeline uses automatic image feature detection
and matching, structure-from-motion and photo-textured terrain surface reconstruction algorithms that require minimal human input
and only a small number of ground control points and allow the use of cheap, consumer-grade digital cameras. The resulting maps
combine colour and topographic information at sub-centimeter resolutions over an area of approximately 100m, thus enabling spatial
properties of the intertidal environment to be determined across a hierarchy of spatial scales. Results of the system are presented for
an intertidal rock platform at Cape Banks, Sydney, Australia. Potential uses of this technique include mapping of plant (micro- and
macro-algae) and animal (e.g. gastropods) assemblages at multiple spatial and temporal scales.
1 INTRODUCTION
Plant and animal assemblages that live in intertidal regions such
as rocky shores are part of a complex, dynamic ecosystem, the
structure and functioning of which can vary across a cascade
of spatial and temporal scales (Underwood, 2000). Intertidal
ecosystems have primarily been studied using field-based sam-
pling (e.g. (Murphy et al., 2008)) at appropriate resolutions to
capture the spatial variability at which assemblages occur. Re-
mote sensing is the ideal tool to collect contiguous data over
large areas in a snapshot of time. Sensors on satellite and aircraft
platforms provide data at relatively coarse spatial (greater than
30cm/pixel) and temporal resolution and this constrains the use-
fulness of these data to test ecological models. Conventional re-
mote sensing systems do not provide information on topographic
variability at small scales (centimeters and meters), which is known
to influence the distribution of assemblages of plant and animal
species. The high costs of data collection further limit their effec-
tiveness in small-scale environmental science and ecology stud-
ies.
In this paper we develop data collection techniques and data pro-
cessing algorithms for constructing ultra-high resolution (sub-
centimeter) three-dimensional (3D) maps of intertidal rock plat-
forms with a low-cost kite-based mapping system using hundreds
of photographs over a broad area. Low-cost platforms such as
kites and balloons are an alternative means for collecting data
over small areas and have been used for aerial photography, for
example in agriculture (Aber et al., 1999, Aber et al., 2002, Arti-
gas and Pechmann, 2010), ecology (Guichard et al., 2000), hu-
manitarian applications (Sklaver et al., 2006) and archaeology
(Verhoeven, 2009) although typically to acquire a single or a
small number of photographs. For example, (Guichard et al.,
2000) developed a stereo photography rig carried by a helium
balloon for collecting stereoscopic image pairs over rocky shores.
243
Our method instead fuses information from hundreds and poten-
tially thousands of overlapping monocular images using modern
photogrammetry and bundle adjustment techniques (Frahm et al.,
2010, Agarwal et al., 2010), allowing for coverage over much
larger areas with high-resolution.
In this paper, we describe a photogrammetric pipeline that was
developed for constructing high-resolution, 3D, photo-realistic
terrain models using multiple low-altitude images collected from
a consumer-grade digital camera on-board the kite platform. The
data processing utilises automatic image feature detection and
matching strategies (Vedaldi and Fulkerson, 2008, Beis and Lowe,
2003) that minimise the number of ground control points required
during data collection and human-effort required in post-processing.
Matched image features were used to estimate the relative pose
between cameras and the 3D positions of image features (up-to an
unknown scale factor). Image pairs were combined incrementally
to form a complete 3D point cloud of the scene and relative cam-
era positions (up-to an unknown scale, rotation and translation)
using structure-from-motion and bundle adjustment (Snavely et
al., 2008). Multi-view stereo reconstruction (Furukawa and Ponce,
2010) was then used to create a dense, 3D pointcloud from which
a triangulated surface model was created. Finally, the surface
model was photo-textured using a band-limited blending of mul-
tiple image patches at each region of the surface and visualised
in 3D using a level-of-detail rendering system to capture sub-
centimeter details over the entire span of the map. The resulting
maps combine colour and topographic information, thus enabling
spatial properties of the intertidal environment to be determined
across a hierarchy of spatial scales. Potential uses of this tech-
nique include mapping of plant (micro- and macro-algae) and an-
imal (e.g. gastropods) assemblages at multiple spatial and tem-
poral scales.
Section 2 describes our approach to kite-based image acquisition
and map processing in detail. Results of the technique are demon-
