Professional Documents
Culture Documents
AAG 2018 Web
AAG 2018 Web
James T. Dietrich*, Department of Geography, University of Northern Iowa, USA james.dietrich@uni.edu py_sfm_depth
Amy S. Woodget, Institute of Science & Environment, University of Worcester, UK https://geojames.github.io/py_sfm_depth/
Refraction Correction One open question with SfM bathymetry is what the SfM process is “seeing” at the bottom
of the river bed. With all the different camera angles used to build the SfM model, each
The underwater portions of the dense point cloud needed to be corrected camera “should” be seeing the river bed at different apparent depths (illustrated right).
for refraction. This was accomplished with py_sfm_depth, a custom python
processing script written by the author and available via Github. The
software calculates all of the To try to answer this question we took the known depths for the 2016 data (363 RTK-GPS
possible combinations of camera/ validation points) and inverted the correction equation to solve for ha given all of the
different camera angles. These are the theoretical apparent depths (hat). Change Conclusions
point combinations, calculates the
refraction angles, and then corrects • The 2016 data had excellent accuracy, but the 2017 data was less accurate.
the bed elevations based on a
Theoretical apparent depths vs. SfM apparent depths One major contributor to the error in the 2017 data is the water surface,
derivation of Snell’s Law: which was much more complex than the 2016 (e.g. more riffle and runs).
• The theoretical apparent depths actually matched with the SfM apparent
depths better that expected. This bodes well for the refraction correction and
give us a good starting point for improvements in the core algorithms and
statistics.
(Dietrich 2017) • Overall, refraction corrected SfM data will be an important tool moving
Trigonometry of the refraction angles for a single camera
forward for geomorphic and river habitat change studies in clear water
systems.
Banks Uncorrected SfM
References
WATER SURFACE Dietrich JT. 2017. Bathymetric Structure-from-Motion: extracting shallow stream bathymetry from multi-view stereo
photogrammetry. Earth Surface Processes and Landforms 42 : 355–364. DOI: 10.1002/esp.4060
Shintani C, Fonstad MA. 2017. Comparing remote-sensing techniques collecting bathymetric data from a gravel-bed river.
International Journal of Remote Sensing 38 : 2883–2902. DOI: 10.1080/01431161.2017.1280636
Woodget AS, Carbonneau PE, Visser F, Maddock IP. 2015. Quantifying submerged fluvial topography using hyperspatial
Refraction Corrected resolution UAS imagery and structure from motion photogrammetry. Earth Surface Processes and Landforms 40 : 47–64.
Point cloud cross-section illustrating the different SfM surfaces DOI: 10.1002/esp.3613