defining and visually modeling salient landscape structures at multiple scales. Our
results support the idea that multiscale analysis (including the establishment of
scale hierarchies) should be guided by the innate spatial resolution of the (salient)
geo-objects composing a scene. Scale space originates from the computer vision
community, where it was developed to analyze real-world structures with no a
priori information about the scene being assessed. Its basic premise is that a
multiscale representation of a signal (such as a remote sensing image of
a landscape) is an ordered set of derived signals showing structures at coarser
scales that constitute simplifications of corresponding structures at finer scales. To
define scale domains we first describe methods that define and link structures at
different scales in scale space to higher order objects, called scale-space blobs, from
which we extract significant features based on their appearance and persistence over
all scales. This is based on the idea that bloblike structures which persist in scale
space are likely candidates to correspond to significant structures in the image and
thus in the landscape.
By integrating concepts from scale space and hierarchy theory:
(i) We describe a three-tier hierarchical methodology for automatically delineating and assessing the dominant structural components within 200 different
multiscale representations of a complex agroforested landscape (with reduced
processing demands).
(ii) We implement a new multiscale geo-object topology and define four new
“species” of dual-instance blob events (Section 8.3.1.3).
(iii) We visually model five different “landscape scale domains” based on the
novel idea of using SS annihilation events as critical domain thresholds.
(iv) We further suggest that the resulting manifolds may be considered a multiscale extension to the hierarchical scaling ladder as defined in the hierarchical
patch dynamics paradigm.
Though beyond the scope of this chapter, we provide ideas for how to implement
these methods over larger scenes and suggest that these domain structures may
represent critical landscape scale thresholds. Thus they may be used as templates to
define the grain and extent over which scale-dependent ecological models could be
developed and applied and the limits over/between which landscape data can be
appropriately scaled.
ACKNOWLEDGMENTS
Dr. Hay completed this chapter while on sabbatical in Brisbane, Australia, at the
Centre for Spatial Environmental Research, School of Geography Planning and
Environmental Management, The University of Queensland, and expresses thanks to
Dr. Stuart Phinn and his team for the collaboration and ideas that have been shared.
Research support has been provided by an NSERC Discovery Grant and Tecterra
ACKNOWLEDGMENTS
163
results support the idea that multiscale analysis (including the establishment of
scale hierarchies) should be guided by the innate spatial resolution of the (salient)
geo-objects composing a scene. Scale space originates from the computer vision
community, where it was developed to analyze real-world structures with no a
priori information about the scene being assessed. Its basic premise is that a
multiscale representation of a signal (such as a remote sensing image of
a landscape) is an ordered set of derived signals showing structures at coarser
scales that constitute simplifications of corresponding structures at finer scales. To
define scale domains we first describe methods that define and link structures at
different scales in scale space to higher order objects, called scale-space blobs, from
which we extract significant features based on their appearance and persistence over
all scales. This is based on the idea that bloblike structures which persist in scale
space are likely candidates to correspond to significant structures in the image and
thus in the landscape.
By integrating concepts from scale space and hierarchy theory:
(i) We describe a three-tier hierarchical methodology for automatically delineating and assessing the dominant structural components within 200 different
multiscale representations of a complex agroforested landscape (with reduced
processing demands).
(ii) We implement a new multiscale geo-object topology and define four new
“species” of dual-instance blob events (Section 8.3.1.3).
(iii) We visually model five different “landscape scale domains” based on the
novel idea of using SS annihilation events as critical domain thresholds.
(iv) We further suggest that the resulting manifolds may be considered a multiscale extension to the hierarchical scaling ladder as defined in the hierarchical
patch dynamics paradigm.
Though beyond the scope of this chapter, we provide ideas for how to implement
these methods over larger scenes and suggest that these domain structures may
represent critical landscape scale thresholds. Thus they may be used as templates to
define the grain and extent over which scale-dependent ecological models could be
developed and applied and the limits over/between which landscape data can be
appropriately scaled.
ACKNOWLEDGMENTS
Dr. Hay completed this chapter while on sabbatical in Brisbane, Australia, at the
Centre for Spatial Environmental Research, School of Geography Planning and
Environmental Management, The University of Queensland, and expresses thanks to
Dr. Stuart Phinn and his team for the collaboration and ideas that have been shared.
Research support has been provided by an NSERC Discovery Grant and Tecterra
ACKNOWLEDGMENTS
163