![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
3-D uncertainty-based topographic change detection with structure-from-motion photogrammetry and precision maps |
VerfasserIn |
Mike R. James, Stuart Robson, Mark W. Smith |
Konferenz |
EGU General Assembly 2017
|
Medientyp |
Artikel
|
Sprache |
en
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 19 (2017) |
Datensatznummer |
250143479
|
Publikation (Nr.) |
EGU/EGU2017-7200.pdf |
|
|
|
Zusammenfassung |
Structure-from-motion (SfM) software greatly facilitates the generation of 3-D surface
models from photographs, but doesn’t provide the detailed error metrics that are
characteristic of rigorous photogrammetry. Here, we present a novel approach to
generate maps of 3-D survey precision which describe the spatial variability in
3-D photogrammetric and georeferencing precision across surveys. Such maps
then enable confidence-bounded quantification of 3-D topographic change that, for
the first time, specifically account for the precision characteristics of photo-based
surveys.
Precision maps for surveys georeferenced either directly using camera positions or by
ground control, illustrate the spatial variability in precision that is associated with the relative
influences of photogrammetric (e.g. image network geometry, tie point quality) and
georeferencing considerations. For common SfM-based software (which does not provide
precision estimates directly), precision maps can be generated using a Monte Carlo
procedure. Confidence-bounded full 3-D change detection between repeat surveys with
associated precision maps, is then derived through adapting a state-of-the-art point-cloud
comparison (M3C2; Lague, et al., 2013).
We demonstrate the approach using annual aerial SfM surveys of an eroding badland,
benchmarked against TLS data for validation. 3-D precision maps enable more probable
erosion patterns to be identified than existing analyses. If precision is limited by weak
georeferencing (e.g. using direct georeferencing with camera positions of multi-metre
precision, such as from a consumer UAV), then overall survey precision scales as n−1 ∕2
of the
control precision (n = number of images). However, direct georeferencing results from SfM
software (PhotoScan) were not consistent with those from rigorous photogrammetric
analysis.
Our method not only enables confidence-bounded 3-D change detection and
uncertainty-based DEM processing, but also provides covariance information for all
parameters. Thus, we now open the door for SfM practitioners to use the comprehensive
analyses that have underpinned rigorous photogrammetric approaches over the last
half-century. |
|
|
|
|
|